lcgwiki - Contributions [en]

Atlas:StorageRequirements

2009-08-05T07:37:16Z

Ueda: /* ATLASDATADISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Total estimation of size comes later

Reprocessing Aug 09
ESD 83TB, AOD 7TB, DPD 13TB, HIST/NTUP/LOG: 5TB, TOTAL 108TB

Fast reprocessing of Cosmic Jun09
ESD 24.5TB
AOD 1.8TB
DPD 6.2TB

Cosmic Jun 09
~300 TB RAW
~100 TB ESD, AOD, DPD

- From first two days, no DPD merging:
Type #files avg size [MB] total x7
------------------------------------
RAW 13278 2020 25.6 TB 180 TB
ESD 10211 240 2.3 TB 16 TB
AOD 586 206 0.12 TB 0.8 TB
DPD 14017 57 0.76 TB 5.3 TB

Spring09 reprocessing
ESD 83TB

* Reprocessing Aug09 (30 July 2009) http://indico.cern.ch/materialDisplay.py?contribId=5&materialId=slides&confId=65114
* Cosmic Jun09 http://indico.cern.ch/getFile.py/access?contribId=8&resId=0&materialId=1&confId=62496

fast reprocessing of june cosmic data
57 runs, 338 datasets, 145660 ﬁles, 295.4TB
https://twiki.cern.ch/twiki/bin/view/Atlas/DataPreparationReprocessing#Fast_reprocessing_of_June_July_2
http://indico.cern.ch/materialDisplay.py?contribId=4&materialId=slides&confId=64069

* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This space has been decomissioned and replaced with ATLASSCRATCHDISK.
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-08-04T15:24:37Z

Ueda: /* ATLASDATADISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Total estimation of size comes later

Reprocessing Aug 09
ESD 83TB, AOD 7TB, DPD 13TB, HIST/NTUP/LOG: 5TB, TOTAL 108TB

Cosmic Jun 09
~300 TB RAW
~100 TB ESD, AOD, DPD

- From first two days, no DPD merging:
Type #files avg size [MB] total x7
------------------------------------
RAW 13278 2020 25.6 TB 180 TB
ESD 10211 240 2.3 TB 16 TB
AOD 586 206 0.12 TB 0.8 TB
DPD 14017 57 0.76 TB 5.3 TB

Spring09 reprocessing
ESD 83TB

* Reprocessing Aug09 (30 July 2009) http://indico.cern.ch/materialDisplay.py?contribId=5&materialId=slides&confId=65114
* Cosmic Jun09 http://indico.cern.ch/getFile.py/access?contribId=8&resId=0&materialId=1&confId=62496

fast reprocessing of june cosmic data
57 runs, 338 datasets, 145660 ﬁles, 295.4TB
https://twiki.cern.ch/twiki/bin/view/Atlas/DataPreparationReprocessing#Fast_reprocessing_of_June_July_2
http://indico.cern.ch/materialDisplay.py?contribId=4&materialId=slides&confId=64069

* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This space has been decomissioned and replaced with ATLASSCRATCHDISK.
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-07-09T08:39:12Z

Ueda: /* ATLASDATADISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Total estimation of size comes later

Cosmic Jun 09
~300 TB RAW
~100 TB ESD, AOD, DPD

- From first two days, no DPD merging:
Type #files avg size [MB] total x7
------------------------------------
RAW 13278 2020 25.6 TB 180 TB
ESD 10211 240 2.3 TB 16 TB
AOD 586 206 0.12 TB 0.8 TB
DPD 14017 57 0.76 TB 5.3 TB

http://indico.cern.ch/getFile.py/access?contribId=8&resId=0&materialId=1&confId=62496

fast reprocessing of june cosmic data
57 runs, 338 datasets, 145660 ﬁles, 295.4TB
https://twiki.cern.ch/twiki/bin/view/Atlas/DataPreparationReprocessing#Fast_reprocessing_of_June_July_2
http://indico.cern.ch/materialDisplay.py?contribId=4&materialId=slides&confId=64069

* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This space has been decomissioned and replaced with ATLASSCRATCHDISK.
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-06-25T15:12:34Z

Ueda: /* ATLASDATADISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Total estimation of size comes later

Cosmic Jun 09
~300 TB RAW
~100 TB ESD, AOD, DPD

- From first two days, no DPD merging:

Type #files avg size [MB] total x7
------------------------------------
RAW 13278 2020 25.6 TB 180 TB
ESD 10211 240 2.3 TB 16 TB
AOD 586 206 0.12 TB 0.8 TB
DPD 14017 57 0.76 TB 5.3 TB

http://indico.cern.ch/getFile.py/access?contribId=8&resId=0&materialId=1&confId=62496

* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This space has been decomissioned and replaced with ATLASSCRATCHDISK.
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-06-22T20:40:48Z

Ueda: /* ATLASUSERDISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This space has been decomissioned and replaced with ATLASSCRATCHDISK.
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-06-19T15:24:03Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The share used for STEP09 http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html [[Image:STEP09-FR-T2-share-used.PNG]]


2009.05.19. (the xls file used to create this table [[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* shares and other assumptions same as the calculation 2009.05.18.

<ul>
<pre>

STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5
</pre>
</ul>


2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18 [[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to http://atladcops.cern.ch:8000/drmon/ftmon_TiersInfo.html (assuming 10% per stream) [[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<ul>
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>
</ul>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

File:STEP09-FR-T2-share-used.PNG

2009-06-19T15:15:58Z

Ueda:

Atlas:StorageRequirements

2009-06-19T15:03:44Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.19. ([[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]])
STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-06-19T15:03:24Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.19. ([[Image:STEP09-FR.xls|STEP09-FR.xls]] [[Image:STEP09-FR-T2-Disk.xls|STEP09-FR-T2-Disk.xls]]
STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

File:STEP09-FR-T2-Disk.xls

2009-06-19T15:00:05Z

Ueda: version 19 Mai 2009

version 19 Mai 2009

Atlas:StorageRequirements

2009-06-19T14:39:15Z

Ueda: /* ATLASGROUPDISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: <strike>/atlas</strike> /atlas/Role=production
** <strike>A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP</strike>
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** <strike>$DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users</strike>
** eg. for phys-higgs</strike>

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.19.
STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5

[[Image:STEP09-FR.xls]]

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-19T20:41:33Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.19.
STEP09 Share CCRC space DATA MCDISK PROD GROUP SCRATCH USER Total LOCALGROUP
112TB 112TB /112TB 30% 25% 5% 20% 20%
IN2P3-CC pledge 0.0
deploy 212.0 454.7 11.0 2.7 4.0 684.4 1.0
free 71.8 78.1 5.7 2.6 2.1 160.3 0.7
IN2P3-LAPP pledge 37.0% 41.4 34.5 6.9 27.6 27.6 0.0 138.0
deploy 20% 25% 6.1% 6.8 14.0 6.0 2.0 1.0 0.5 30.3 3.6
free 22.4 28.0 4.0% 4.5 0.7 5.0 2.0 0.9 0.4 13.5 2.1
IN2P3-LPC pledge 38.8% 43.5 36.3 7.3 29.0 29.0 0.0 145.0
deploy 30% 25% 4.5% 5.0 17.0 3.0 2.0 3.0 3.0 33.0 1.0
free 33.6 28.0 4.0% 4.5 12.8 2.6 2.0 2.2 3.0 27.1 0.9
IN2P3-CPPM capacity 4.3% 4.9 4.1 0.8 3.2 3.2 0.0 16.2
deploy 20% 5% 2.6% 2.9 8.8 2.9 0.1 1.0 0.5 16.2 0.1
free 22.4 5.6 0.8% 0.9 1.1 2.5 0.1 0.9 0.5 6.0 0.1
IN2P3-LPSC capacity 3.9% 4.4 3.6 0.7 2.9 2.9 0.0 14.5
deploy 5% 5% 0.9% 1.0 10.0 2.0 0.5 0.5 0.5 14.5
free 5.6 5.6 0.6% 0.7 6.9 1.5 0.5 0.5 0.4 10.5
BEIJING-LCG2 pledge 53.6% 60.0 50.0 10.0 40.0 40.0 0.0 200.0
deploy 12% 20% 4.5% 5.0 25.0 2.0 1.0 5.0 1.0 39.0
free 13.4 22.4 4.4% 4.9 14.2 1.6 1.0 5.0 1.0 27.7
GRIF pledge 182.4% 204.3 170.3 34.1 136.2 136.2 0.0 681.0
deploy 74.1% 83.0 87.2 24.0 22.0 9.6 1.0 226.8 22.0
free 66.0% 73.9 41.9 23.1 22.0 9.5 1.0 171.4 21.9
GRIF-LAL deploy 45% 45% 8.9% 10.0 30.0 6.0 6.0 3.0 55.0 2.0
free 50.4 50.4 2.5% 2.8 18.6 5.5 6.0 3.0 35.9 1.9
GRIF-SACLAY deploy 30% 30% 3.6% 4.0 29.2 2.0 2.6 0.0 37.8
free 33.6 33.6 3.6% 4.0 5.4 1.7 2.6 0.0 13.7
GRIF-LPNHE deploy 25% 25% 61.6% 69.0 28.0 16.0 16.0 4.0 1.0 134.0 20.0
free 28.0 28.0 59.9% 67.1 17.9 15.9 16.0 3.9 1.0 121.8 20.0
TOKYO-LCG2 pledge 107.1% 120.0 100.0 20.0 80.0 80.0 0.0 400.0
deploy 100% 100% 121.8% 136.4 109.1 13.6 9.1 9.1 9.1 286.4 9.1
free 112.0 112.0 107.3% 120.2 17.6 11.0 9.1 8.5 9.0 175.4 2.6
RO federation pledge 51.4% 57.6 48.0 9.6 38.4 38.4 0.0 192.0
deploy 7.1% 8.0 20.0 8.0 6.0 10.0 7.9 59.9 0.0
free 3.5% 3.9 4.2 7.0 6.0 10.0 7.9 39.0 0.0
RO-02-NIPNE deploy 10% 10% 4.5% 5.0 11.0 6.0 3.0 5.0 3.0 33.0
free 11.2 11.2 2.6% 2.9 4.0 5.2 3.0 5.0 3.0 23.1
RO-07-NIPNE deploy 10% 10% 2.7% 3.0 9.0 2.0 3.0 5.0 4.9 26.9
free 11.2 11.2 0.9% 1.0 0.2 1.8 3.0 5.0 4.9 15.9

T2 total pledge 0.0 0.0 470.4% 526.8 439.0 87.8 351.2 351.2 0.0 1756.0 0.0
deploy 307.0% 300.0% 302.8% 339.1 398.3 93.5 70.7 58.8 32.4 992.8 57.8
free 343.8 336.0 260.1% 291.3 145.5 84.4 70.7 57.0 32.1 681.0 49.5

[[Image:STEP09-FR.xls]]

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

File:STEP09-FR.xls

2009-05-19T20:40:09Z

Ueda:

Atlas:StorageRequirements

2009-05-19T20:19:52Z

Ueda: /* STEP09 requirements */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| MC input for analysis
| output of analysis
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
| 50TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T15:21:25Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* usage of MCDISK in STEP09 much smaller than that of DATADISK
* usage of MCDISK before STEP09 unknown
* usage of SCRATCHDISK (especially by user analysis jobs) unknown
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T15:18:59Z

Ueda: /* Introduction */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) [http://indico.cern.ch/conferenceDisplay.py?confId=38738]

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) [http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808]
** LCG-France T2-T3 Technical meeting (18 July 2008) [http://indico.in2p3.fr/conferenceDisplay.py?confId=809]

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T15:18:07Z

Ueda: /* Introduction */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)] and [http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T15:16:56Z

Ueda: /* Introduction */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* https://twiki.cern.ch/twiki/bin/view/Atlas/Step09
* Presentations at [[http://indico.cern.ch/conferenceDisplay.py?confId=45475&view=standard&showDate=all&showSession=all&detailLevel=contribution GDB (13 May 2009)]] and [[http://indico.in2p3.fr/contributionDisplay.py?contribId=0&sessionId=3&confId=1660 Réunion LCG-France (18-19 May 2009)]]
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T15:07:15Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18
[[Image:STEP09-FR-20090518a.PNG]]
* share according to (assuming 10% per stream)
[[Image:STEP09-FR-share-20090518.PNG]]

* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

Old calculation

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

File:STEP09-FR-share-20090518.PNG

2009-05-18T15:03:59Z

Ueda:

File:STEP09-FR-20090518a.PNG

2009-05-18T14:58:40Z

Ueda:

Atlas:StorageRequirements

2009-05-18T14:46:35Z

Ueda: /* STEP09 requirements */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
<ul>
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}
</ul>

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

File:STEP09-FR-20090518.PNG

2009-05-18T14:45:45Z

Ueda: Space requirements for STEP09

Space requirements for STEP09

Atlas:StorageRequirements

2009-05-18T14:44:56Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

2009.05.18.
* pledged capacity [[http://lcg.web.cern.ch/LCG/Resources/WLCGResources-2008-2013-05FEB2009.pdf Tier 1 & Tier 2 Pledge and Resources table (v 18 March 2009)]]
* deployed and free space as of 2009 Mai 18

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T13:25:48Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

2009.05.18.
{|
| Site
|
| Share
| DATADISK
| MCDISK
| PROD
| GROUP
| USER
| SCRATCH
| LOCALGROUP
|-----
| IN2P3-CC
| pledge
| 100%
|-----
|
| total
|
| 222
| 427
| -
| 1.0
| 4.0
| -
| 1.0
|-----
|
| free
|-----
|under construction
|}

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T13:24:46Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

2009.05.18.
{|
| Site
|
| Share
| DATADISK
| MCDISK
| PROD
| GROUP
| USER
| SCRATCH
| LOCALGROUP
|-----
| IN2P3-CC
| pledge
| 100%
|-----
|
| total
| 222
| 427
| -
| 1.0
| 4.0
| -
| 1.0
|-----
| IN2P3-CC
| free
|-----
|under construction
|}

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T13:14:13Z

Ueda: /* STEP09 requirements */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

* T2 Space Token Summary
{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-05-18T13:13:13Z

Ueda: /* Summary table for Space token and Name space */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

=== STEP09 requirements ===
presentation at [[http://indico.cern.ch/materialDisplay.py?contribId=5&sessionId=3&materialId=slides&confId=45475|GDB 13 May 2009]]

* Total data export T1-T2(for 100%): 112 TB, 92.6MB/s

<ul>
{| border="1" width="100%"
|
| T0 AOD+DPD (0.2MB/event each)
| reprocessing AOD+DPD
| data total
| MC prod
| MC repro
| comment
|-----
| 200Hz trigger
| 46.3MB/s (average)
| 46.3MB/s
| 92.6MB/s
|-----
| 50ksec /day
| 4TB/day
| 4TB/day
| 8TB/day
|-----
| STEP09 (2 weeks)
| 56TB
| 56TB
| 112TB
| 3.5TB
| 3.5TB
|}
</ul>

{| border="1" width="100%"
| space token
| space fraction
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| 30%
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| 25%
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| 5%
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| 20%
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]]
| 20%
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]]
| no requirement
|}

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-16T10:36:07Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-16T10:35:35Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB VO:atlas
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB VO:atlas
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbat05.nipne.ro
| 5.0 TB VOMS:/atlas/Role=production
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| Need to correct the group for the space token
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-16T10:30:20Z

Ueda: /* Space Reservation deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* Automatic Query: SRM2.2 space token descriptions advertised in information system and associated ACLs[http://wn3.epcc.ed.ac.uk/srm/xml/srm_token_acls_table?token=ATLASSCRATCHDISK&vo=.*&site=.*&acl=.*]

* The table below shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbat05.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T09:47:05Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| ccsrm.ihep.ac.cn
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbit00.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbat05.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T09:33:27Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbat05.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T09:24:19Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| tbat05.nipne.ro
| 5.0 TB
| root
| root
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T01:16:07Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.977 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T01:14:54Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| 0.488 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T01:11:27Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-11T01:10:04Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| 3.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T17:18:19Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| space token needed, namespace ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T13:40:41Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| I Ueda
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| marsedpm.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx g:atlas:rwx
| d:g::rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas:rwx
| space token needed, namespace ok
|-----
| grid05.lal.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| space token needed, namespace ok
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas
| g::rwx
| d:g::rwx
| need to set at least d:g:atlas:rwx and d:g:atlas/Role=production:rwx
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx g:atlas:rwx g:atlas/Role=production:rwx g:atlas/Role=lcgadmin:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=production:rwx d:g:atlas/Role=lcgadmin:rwx
| ok
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx g:atlas:rwx g:atlas/Role=lcgadmin:rwx g:atlas/Role=production:rwx
| d:g::rwx d:g:atlas:rwx d:g:atlas/Role=lcgadmin:rwx d:g:atlas/Role=production:rwx
| ok
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T11:05:51Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| N/A
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| N/A
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| srm.ihep.ac.cn
| N/A
|-----
| tbit00.nipne.ro
| N/A
|-----
| tbat05.nipne.ro
| N/A
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T11:03:35Z

Ueda: /* ATLASSCRATCHDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
Requirements: [[#ATLASSCRATCHDISK]]
{| border="1" width="100%"
| SE
| Space Reservation
| owner
| group
| g:
| d:g:
| comment
|-----
| ccsrm.in2p3.fr
| NONE
|-----
| lapp-se01.in2p3.fr
| 1.0 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| 3.0 TB
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| 0.5 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| NONE
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| NONE
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| 0.3 TB
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| 4.0 TB
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| 9.1 TB
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| srm.ihep.ac.cn
| NONE
|-----
| tbit00.nipne.ro
| NONE
|-----
| tbat05.nipne.ro
| NONE
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T10:57:09Z

Ueda: /* ATLASUSERDISK deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

== ATLASSCRATCHDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| d:g:
| comment
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| srm.ihep.ac.cn
|-----
| tbit00.nipne.ro
|-----
| tbat05.nipne.ro
|}

== ATLASLOCALGROUPDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-06T10:56:11Z

Ueda: /* Deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
== Space Reservation deployment ==
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08 and old requirements, thus they need to be revisited with the real size.
<pre>
sizes in TB.
Site Share DATA MC PROD GROUP USER SCRATCH LOCALGROUP
IN2P3-CC 100% 222/ 427/ -/- 1.0/ 4.0 - 1.0
IN2P3-LAPP 25% 6.8/ 14/15 6/2.6 2.0/7.5 0.5/ 1.0/ 3.6
IN2P3-LPC 25% 5.0/ 11/15 3/2.4 2.0/7.5 3.0/ 3.0/ 1.0
IN2P3-CPPM 5% 2.9/ 7.8/3 2.9/ 0.1/1.5 0.5/ - 0.1
IN2P3-LPSC 5% 1.0/ 10/3 2/ 0.5/1.5 0.5/ 0.5/ ---
BEIJING-LCG2 20% 4.5/ 9.1/12 0.9/1.2 0.9/6 0.9/ - ---
GRIF-LAL 45% 6/ 15/27 6/ 6/13.5 2/ - 2.0
GRIF-SACLAY 30% 4/ 13.8/18 2/ -/9 2/ 0.3/ ---
GRIF-LPNHE 25% 93/ 23/15 16/ 16/7.5 4/ 4/ 34.0
TOKYO-LCG2 100% 136/ 82/60 5.5/6 9.1/30 9.1/ 9.1/ 9.1
RO-02-NIPNE 10% 5/ 9/6 6/ 3/3 3/ - ---
RO-07-NIPNE 10% 3/ 9/6 2/ 3/3 5/ - ---
</pre>

== ATLASUSERDISK deployment ==
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

; ATLASSCRATCHDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| d:g:
| comment
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| srm.ihep.ac.cn
|-----
| tbit00.nipne.ro
|-----
| tbat05.nipne.ro
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-05T16:24:08Z

Ueda: /* Deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

; ATLASSCRATCHDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| d:g:
| subdirectories
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| srm.ihep.ac.cn
|-----
| tbit00.nipne.ro
|-----
| tbat05.nipne.ro
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-05T16:13:23Z

Ueda: /* Deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

; ATLASSCRATCHDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| d:g:
| subdirectories
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lpsc-se-dpm-server.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas:rwx
| d:g::rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas:rwx
|
|-----
| grid05.lal.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| node12.datagrid.cea.fr
| root
| atlas
| g::rwx
| d:g::rwx
|
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| g::rwx, g:atlas:rwx, g:atlas/Role=production:rwx, g:atlas/Role=lcgadmin:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=production:rwx, d:g:atlas/Role=lcgadmin:rwx
|
|-----
| lcg-se01.icepp.jp
| root
| atlas
| g::rwx, g:atlas:rwx, g:atlas/Role=lcgadmin:rwx, g:atlas/Role=production:rwx
| d:g::rwx, d:g:atlas:rwx, d:g:atlas/Role=lcgadmin:rwx, d:g:atlas/Role=production:rwx
|
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-05T15:23:29Z

Ueda: /* Summary table for Space token and Name space */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASSCRATCHDISK|ATLASSCRATCHDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASSCRATCHDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASSCRATCHDISK|see below*]])
| '''<code>.../atlas/atlasscratchdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2009-03-05T15:21:47Z

Ueda: /* ATLASUSERDISK */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASSCRATCHDISK ==
This space is to be deployed to replace ATLASUSERDISK.
During the migraiton period, the both spaces should be available. The size of the space
is preferably the same as ATLASUSERDISK, but it is also possible to start with a small size
and increase it while monitoring the usage and decreasing ATLASUSERDISK.

This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasscratchdisk/
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasscratchdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasscratchdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2008-09-23T13:34:57Z

Ueda: /* Deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:?, user08.FredericDerue:?, user08.JohannesElmsheuser:?
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}

Atlas:StorageRequirements

2008-09-23T13:12:48Z

Ueda: /* Deployment */

= Introduction =
Sources of information:
* Now ATLAS has prepared a page https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp
* Presentations at ATLAS Tier-1/2/3 Jamboree (28 August 2008) http://indico.cern.ch/conferenceDisplay.py?confId=38738

Some comments about the contents

* Information only for T2s (and T3s).
* ACL information only for DPM for the moment.
* See the slides
** LCG-France T2-T3 Technical meeting (20 June 2008) http://indico.in2p3.fr/materialDisplay.py?contribId=4&materialId=slides&confId=808
** LCG-France T2-T3 Technical meeting (18 July 2008) http://indico.in2p3.fr/conferenceDisplay.py?confId=809

* The requirements for ATLASGROUPDISK is temporary due to a technical limitation in DPM. Atlas has contacted DPM developpers. The proper implementation is expected to be available in September.

* Site admin's are requested to create at least the space tokens. The namespace directories can be left to atlas production if the top directory is properly set up so that <code>/atlas/Role=production</code> is allowed to execute dpns-mkdir and dpns-setacl remotely. (Of course they are welcome to create them by themselves).

* In general ''read'' permission should be given to all ATLAS users everywhere.
* In general ''write'' permission should be given to <code>/atlas/Role=production</code> everywhere, so that the ATLAS central deletion tool can work.
* It is suggested at LAPP to give ''write'' permission to <code>atlas/Role=lcgadmin</code> everywhere, so that the ATLAS admin can control the storage. (less people in <code>Role=lcgadmin</code> than in than in <code>/atlas/Role=production</code>). Stephane will discuss this issue with ATLAS.

= Space Requirements =
== ATLAS top directory ==
* On this page the top directory for ATLAS is represented by <code>.../atlas</code>. Each site should replace it according to their SE configuration.
** eg. <code>lapp-se01.in2p3.fr:/dpm/in2p3.fr/home/atlas</code> for lapp.

* Although it is up to sites policies, but it is recommended
** to have this top directory configured so that only /atlas/Role=production and /atlas/Role=lcgadmin can create files/directories underneath and ordinary users cannot.

== Summary table for Space token and Name space ==

* The size in < > is for a typical T2 with ~500 CPUs and ~100 TB disk

{| border="1" width="100%"
| space token
| voms group to be associated
| space to be reserved
| namespace directory(-ies) to be created
|-----
| [[#ATLASDATADISK|ATLASDATADISK]]
| '''<code>/atlas/Role=production</code>'''
| not known yet [[#ATLASDATADISK|see below*]]
| '''<code>.../atlas/atlasdatadisk</code>'''
write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASMCDISK|ATLASMCDISK]]
| '''<code>/atlas/Role=production</code>'''
| a share of 60TB (plus possible extra) [[#ATLASMCDISK|see below*]]
| '''<code>.../atlas/atlasmcdisk</code>''' write permission only to <code>atlas/Role=production</code>
|-----
| [[#ATLASPRODDISK|ATLASPRODDISK]]
| '''<code>/atlas/Role=production</code>'''
| <2TB>, (or 6GB/kSI2K) ([[#ATLASPRODDISK|see below*]])
| '''<code>.../atlas/atlasproddisk</code>''' write permission only to <code>/atlas/Role=production</code>
|-----
| [[#ATLASGROUPDISK|ATLASGROUPDISK]]
| '''<code>/atlas</code>''' (temporary solution. [[#ATLASGROUPDISK|see below*]])
| a share of 30TB ([[#ATLASGROUPDISK|see below*]])
| '''<code>.../atlas/atlasgroupdisk</code>''' write permission to <code>/atlas/Role=production</code>
 '''<code>.../atlas/atlasgroupdisk/$GROUP</code>''' write permission to <code>/atlas/Role=production</code> and <code>/atlas/$GROUP/Role=production</code>
 ($GROUP: [[#ATLASGROUPDISK|see below*]])
|-----
| [[#ATLASUSERDISK|ATLASUSERDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASUSERDISK_space_token Atlas StorageSetUp])
| '''<code>/atlas</code>'''
| <5TB> ([[#ATLASUSERDISK|see below*]])
| '''<code>.../atlas/atlasuserdisk</code>''' write permission to all ATLAS users write permission to <code>/atlas/Role=production</code> for central deletion the same for all the subdirectories
|-----
| [[#ATLASLOCALGROUPDISK|ATLASLOCALGROUPDISK]] ([https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp])
| '''<code>/atlas/<locality></code>''', <locality>=fr,ro,cn,jp,... '''<code>/atlas</code>'''
| sites to decide
| '''<code>.../atlas/atlaslocalgroupdisk</code>''' write permission to local ATLAS users (<code>/atlas/<locality></code>) write permission to <code>/atlas/Role=production</code> for central deletion
|}

* A T3 will need ATLASDATADISK if it would like to receive real data, ATLASMCDISK to received simulated data, ATLASGROUPDISK to receive group analysis data, and ATLASPRODDISK to contribute official production. ATLASUSERDISK is not necessary, but may be needed if the site contributes non-local user analysis.

* The spacetokens ATLASGRP<GROUP> (eg. ATLASGRPTOP), which were requested for ccrc08/fdr, are not required any longer, being repleaced with a single spacetoken ATLASGROUPDISK. There should not be a file using this spacetoken. They can be removced.

* The spacetoken ATLASENDUSER, which was requested for ccrc08/fdr, is not required any longer, being replaced with ATLASUSERDISK. It can be removed.

== ATLASDATADISK ==
* Estimation of size comes later
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasdatadisk
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>

== ATLASMCDISK ==
This is the space where simulated data produced centrally go in.
* 15TB for a typical T2 with ~500 CPU?fs and ~100 TB disk
** 60TB for a T2 requesting for 100% AOD.
** The size will increase in case D1PD are requested by the site. (100% D1PD will be 30TB)
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space at
Site Share Jul08 Sep08 Dec08 Possibly with D1PD
IN2P3-LAPP 25% 5 TB 10 TB 15 TB 22.5 TB
IN2P3-CPPM 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPSC 5% 1 TB 2 TB 3 TB 4.5 TB
IN2P3-LPC 25% 5 TB 10 TB 15 TB 22.5 TB
BEIJING-LCG2 20% 4 TB 8 TB 12 TB 18 TB
RO-07-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
RO-02-NIPNE 10% 2 TB 4 TB 6 TB 9 TB
GRIF-LAL 45% 9 TB 18 TB 27 TB 40.5 TB
GRIF-LPNHE 25% 5 TB 10 TB 15 TB 22.5 TB
GRIF-SACLAY 30% 6 TB 12 TB 18 TB 27 TB
TOKYO-LCG2 100% 20 TB 40 TB 60 TB 90 TB
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: <code>$DPNS_HOME/atlas/atlasmcdisk</code>
** Normally, sites have already this namespace created.
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:r-x
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASPRODDISK ==
This is the space used by the ATLAS production system (panda) for putting input files for the jobs to run on the site, downloading them from T1, and output files from the jobs which are to be replicated to the T1. Both input and output files are to be deleted centrally after a certain period.
* 2TB for a typical T2 with ~500 CPUs (the size is to be re-visited)
** scales with the CPU capacity of the site.
** will be larger if the reconstruction jobs run on the site.
** My private estimation is 6GB/kSI2K (400kSI2K/event for simulation, 4MB/HITS, 7days to keep).
** Using the MoU value, the required space for each site is;
<ul><ul>
<pre>
Site kSI2K Space
BEIJING 200 1.2 TB
GRIF 800 4.8 TB (federation)
LAPP 440 2.6 TB
LPC 400 2.4 TB
TOKYO 1000 6 TB
RO 400 2.4 TB (federation)
</pre>
</ul></ul>
* VOMS group associated with the space: /atlas/Role=production
* Namespace directory to be created: $DPNS_HOME/atlas/atlasproddisk
* Namespace ACL: writable by only atlas/Role=production, readable by all ATLAS users
<ul>
<pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASGROUPDISK ==
This is the space to put group DPD files, which are not created by the central production system, but by group analyses with the group production roles.
* 6TB for a typical T2 with ~500 CPUs and ~100 TB disk
** 100% D2PD will be 30TB
** The required space is 1/3 of the above for the period until Sep 2008, 2/3 for Sep - Dec 2008, and full for Jan - Mar 2009.
** using the share at ccrc08, the required size for each FR sites are;
<ul><ul>
<pre>
Required Space by
Site Share Jul08 Sep08 Dec08
IN2P3-LAPP 25% 2.5 TB 5 TB 7.5 TB
IN2P3-CPPM 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPSC 5% 0.5 TB 1 TB 1.5 TB
IN2P3-LPC 25% 2.5 TB 5 TB 7.5 TB
BEIJING-LCG2 20% 2 TB 4 TB 6 TB
RO-07-NIPNE 10% 1 TB 2 TB 3 TB
RO-02-NIPNE 10% 1 TB 2 TB 3 TB
GRIF-LAL 45% 4.5 TB 9 TB 13.5 TB
GRIF-LPNHE 25% 2.5 TB 5 TB 7.5 TB
GRIF-SACLAY 30% 3 TB 6 TB 9 TB
TOKYO-LCG2 100% 10 TB 20 TB 30 TB
</pre>
</ul></ul>
* One single space to be reserved for all the group activities ($GROUP = phys-beauty, phys-exotics, phys-gener, phys-hi, phys-higgs, phys-lumin, phys-sm, phys-susy, phys-top, perf-egamma, perf-flavtag, perf-jets, perf-muons, perf-tau, etc.)
* VOMS group associated with the space: /atlas
** A temporary solution until multiple group support to the spaces is available.
** Once it is available, the groups will be /atlas/Role=production and /atlas/$GROUP/Role=production for all $GROUP
* Namespaces directory to be created and their ACLs:
** $DPNS_HOME/atlas/atlasgroupdisk: writable by only atlas/Role=production, readable by all ATLAS users
** $DPNS_HOME/atlas/atlasgroupdisk/$GROUP: writable by atlas/Role=production and /atlas/$GROUP/Role=production, readable by all ATLAS users
** eg. for phys-higgs

<ul><pre>
# group: atlas/Role=production
user::rwx
group::rwx #effective:rwx
group:atlas/Role=production:rwx #effective:rwx
group:atlas/phys-higgs/Role=production:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas/Role=production:rwx
default:group:atlas/phys-higgs/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre>
</ul>

== ATLASUSERDISK ==
This is the scratch space for users to put the output of their analysis jobs run at the site before copying them to their final destination. The files are to be deleted by central operation after a certain period of time.
* 5TB for a typical T2 with ~500 CPUs and ~100 TB disk
* VOMS group associated with the space: /atlas
* Namespace directory to be created: .../atlas/atlasuserdisk/ (the old .../atlas/user is to be used for files without spacetoken)
* Namespace ACL: writable by all ATLAS users (group /atlas) and /atlas/Role=production (for central deletion). <strike>The user directories underneath should be writable only by the owner and /atlas/Role=production (for central deletion)</strike>. all the subdirectories should follow this.
* Normally, sites have already this namespace created by user analysis jobs so far.
* example commands
** <code>dpns-entergrpmap --group "atlas"</code>
** <code>dpns-entergrpmap --group "atlas/Role=production"</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-mkdir $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas:rwx,m:rwx,d:g:atlas:r-x,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-setacl -m g:atlas/Role=production:rwx,m:rwx,d:g:atlas/Role=production:rwx,d:m:rwx $DPNS_HOME/atlas/atlasuserdisk</code>
** <code>dpns-getacl $DPNS_HOME/atlas/atlasuserdisk</code>
<ul><pre>
# group: atlas
user::rwx
group::rwx
group:atlas:rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::rwx
default:group:atlas:rwx
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

== ATLASLOCALGROUPDISK ==
This is a space for "local" users of the site, and is not included in the pledged resources to ATLAS.
* See the [https://twiki.cern.ch/twiki/bin/view/Atlas/StorageSetUp#The_ATLASLOCALGROUPDISK_space_to Atlas StorageSetUp] wiki page.
* size to be decided by sites.
** the resources not included in the pledge.
* VOMS group associated with the space: <strike>/atlas/fr (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly)</strike> /atlas
* name space: $DPNS_HOME/atlas/atlaslocalgroupdisk (the path previously required .../atlas/fr/user (or .../atlas/ro/user, .../atlas/cn/user, .../atlas/jp/user, etc.) is to be used for files without space token)
* ACL: write permission <strike>only</strike> to /atlas/fr group (or /atlas/ro, /atlas/cn, /atlas/jp correspondingly) and /atlas/Role=production
* example ACL:
<ul><pre>
# group: atlas/fr
user::rwx
group::rwx
group:atlas/Role=lcgadmin:rwx
group:atlas/Role=production:rwx
group:atlas/fr:rwx #effective:rwx
mask::rwx
other::r-x
default:user::rwx
default:group::r-x
default:group:atlas:r-x
default:group:atlas/Role=lcgadmin:rwx
default:group:atlas/Role=production:rwx
default:group:atlas/fr:rwx
default:mask::rwx
default:other::r-x
</pre></ul>

= Deployment =
; Space Reservation :
* The table shows spaces '''reserved/required'''.
** The reserved spaces are obtained using srm commands.
** The required spaces are obtained using the shares at ccrc08, thus they might change.
<pre>
Site Share DATADISK MCDISK PRODDISK GROUPDISK USERDISK LOCAL
IN2P3-LAPP 25% 5/ TB 5/15 TB 2/2.6 0.5/7.5 TB 0.5/ 0.5 TB
IN2P3-LPC 25% 5/ TB 5/15 TB 2/2.4 1/7.5 TB 2/
IN2P3-CPPM 5% 1/ TB 1.5/3 TB 1/ 0.5/1.5 TB 0.5/ TB (0.5) TB
IN2P3-LPSC 5% 1/ TB 2/3 TB 0.5/ 0.5/1.5 TB 0.5/
BEIJING-LCG2 20% 4.5/ TB 5.5/12 TB 0.9/1.2 0.9/6 TB 0.9/
RO-07-NIPNE 10% 3/ TB 9/6 TB 2/TB 3/3 TB 5/TB
RO-02-NIPNE 10% 5/ TB 9/6 TB 2/ 3/3 TB 3/
GRIF-LAL 45% 6/ TB 15/27 TB 6/ 6/13.5 TB 2/
GRIF-SACLAY 30% 4/ 6/18 TB 2/ /9 TB 2/
GRIF-LPNHE 25% 4/ TB 10/15 TB 1/ 4/7.5 TB 3/ 1.0 TB
TOKYO-LCG2 100% 15/ TB 15/60 TB 2/6 TB /30 TB 5/ 5.0 TB
</pre>

; ATLASUSERDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g:
| d:g:atlas
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| atlas
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user:ok, user08:ok, user08.EricLancon:NG, user08.FredericDerue:NG, user08.JohannesElmsheuser:NG
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
| SAM:ok, user:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:ok
|-----
| lpnse1.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08:ok, user08.EricLancon:ok, user08.FredericDerue:ok, user08.JohannesElmsheuser:
|-----
| lapp-se01.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, users:ok, user08:ok, user08.EricLancon:?, user08.JohannesElmsheuser:?
|-----
| marsedpm.in2p3.fr
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| no subdirectory
|-----
| lpsc-se-dpm-server.in2p3.fr
| Alessandro Di Girolamo
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok
|-----
| lcg-se01.icepp.jp
| I Ueda
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| not checked yet
|-----
| tbat05.nipne.ro
| root
| atlas/Role=production
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| SAM:ok, user08.EricLancon:?
|-----
| tbit00.nipne.ro
| Eric Lancon
| atlas/Role=pilot
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| rwx
| user08.EricLancon:?
|}

;ATLASLOCALGROUPDISK:
{| border="1" width="100%"
| SE
| owner
| group
| g:
| g:atlas
| g:atlas/ <locality>
| g:atlas/ Role= production
| g:atlas/ Role= lcgadmin
| d:g
| d:g:atlas
| d:g:atlas/ <locality>
| d:g:atlas/ Role= production
| d:g:atlas/ Role= lcgadmin
| subdirectories
|-----
| clrlcgse01.in2p3.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| grid05.lal.in2p3.fr
| Stephane Jezequel
| atlas/fr
| rwx
| r-x
| none
| rwx
| rwx
| rwx
| '''r-x'''
| rwx
| rwx
| rwx
| no subdirectory, 1 file (test1 by Stephane)
|-----
| node12.datagrid.cea.fr
| root
| root
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| jerome:NG
|-----
| lpnse1.in2p3.fr
| root
| atlas/ Role=production
| rwx
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| none
|-----
| lapp-se01.in2p3.fr
| root
| atlas/ Role=production
| r-x
| none
| rwx
| rwx
| rwx
| rwx
| none
| rwx
| rwx
| rwx
| SAM:ok
|-----
| marsedpm.in2p3.fr
|-----
| lpsc-se-dpm-server.in2p3.fr
|-----
| lcg-se01.icepp.jp
| root
| atlas/jp
| rwx
| none
| rwx
| rwx
| rwx
| '''r-x'''
| none
| rwx
| rwx
| rwx
| SAM:ok, data08_cos:?, fdr08_run2:?
|-----
| tbat05.nipne.ro
|-----
| tbit00.nipne.ro
|}