Difference between revisions of "Atlas:Analysis ST 2009"

Un article de lcgwiki.
Jump to: navigation, search
(FR Cloud ST (2009 plans))
(FR Cloud ST (2009 plans))
Ligne 2: Ligne 2:
 
=== FR Cloud ST (2009 plans) ===
 
=== FR Cloud ST (2009 plans) ===
 
* [http://lcg.in2p3.fr/wiki/index.php/Atlas:Analysis_Challenge_ST  Details of Site Stress test] : procedure, test conditions and targets
 
* [http://lcg.in2p3.fr/wiki/index.php/Atlas:Analysis_Challenge_ST  Details of Site Stress test] : procedure, test conditions and targets
* '''Weekly ST''' : 50 jobs per site - all sites involved - ''MC Production : FR Cloud ON'''
+
* '''Weekly ST''' : 50 jobs per site - all sites involved - ''MC Production : FR Cloud ON''
 
** LPSC : included (tested during http://gangarobot.cern.ch/st/test_117/)
 
** LPSC : included (tested during http://gangarobot.cern.ch/st/test_117/)
 
** BEIJING : switch to SE dpm ccsrm.ihep.ac.cn (tested during http://gangarobot.cern.ch/st/test_117/)- VO_ATLAS_DEFAULT_SE value (srm.ihep.ac.cn) to be changed, should be OK now
 
** BEIJING : switch to SE dpm ccsrm.ihep.ac.cn (tested during http://gangarobot.cern.ch/st/test_117/)- VO_ATLAS_DEFAULT_SE value (srm.ihep.ac.cn) to be changed, should be OK now
 
** RO-02-NIPNE : contact e-mail atgrid@ifin.nipne.ro, Site admins : Gabriel Stoicea & Serban Constantinescu
 
** RO-02-NIPNE : contact e-mail atgrid@ifin.nipne.ro, Site admins : Gabriel Stoicea & Serban Constantinescu
* '''File access comparison''' : 2009-01-29 / 2009-01-31 ''MC Production : FR Cloud ON'''
+
* '''File access comparison''' : 2009-01-29 / 2009-01-31 ''MC Production : FR Cloud ON''
 
**http://gangarobot.cern.ch/st/test_124/  50 jobs - Posix I/O - Start Time 16:00 UTC  
 
**http://gangarobot.cern.ch/st/test_124/  50 jobs - Posix I/O - Start Time 16:00 UTC  
 
**http://gangarobot.cern.ch/st/test_125/  50 jobs - FileStager - Start Time 16:30 UTC  
 
**http://gangarobot.cern.ch/st/test_125/  50 jobs - FileStager - Start Time 16:30 UTC  
* '''Optimization and Site Stress test''': on demand and dedicated to sites which are interrested - ''MC Production : FR Cloud OFF'''
+
* '''Optimization and Site Stress test''': on demand and dedicated to sites which are interrested - ''MC Production : FR Cloud OFF''
 
** Feb 4-5 (in preparation - to be confirmed)
 
** Feb 4-5 (in preparation - to be confirmed)
 
  GRIF : GRIFOPN test - Total of 1000 jobs will
 
  GRIF : GRIFOPN test - Total of 1000 jobs will

Version du 12:35, 29 janvier 2009

29.01.09

FR Cloud ST (2009 plans)

  • Details of Site Stress test : procedure, test conditions and targets
  • Weekly ST : 50 jobs per site - all sites involved - MC Production : FR Cloud ON
  • File access comparison : 2009-01-29 / 2009-01-31 MC Production : FR Cloud ON
  • Optimization and Site Stress test: on demand and dedicated to sites which are interrested - MC Production : FR Cloud OFF
    • Feb 4-5 (in preparation - to be confirmed)
GRIF : GRIFOPN test - Total of 1000 jobs will
be distributed to ATLAS CEs with following DS available being accessed
from LAL, SACLAY and LPNHE
       - GRIF-SACLAY_MCDISK: 332
       - GRIF-LPNHE_MCDISK: 334
       - GRIF-LAL_MCDISK: 457
LAPP-CPPM : interrested by a stress test ~200-300 jobs 
LPSC : interrested by a stress test ~200-300 jobs

Input DS available

  • Thanks to Stéphane, followong DS available at sites (27/01/09):
{'TOKYO-LCG2_MCDISK': 1038, 'IN2P3-LPC_MCDISK': 342,
'IN2P3-CPPM_MCDISK': 185, 'GRIF-SACLAY_MCDISK': 369,
'GRIF-LPNHE_MCDISK': 380, 'IN2P3-LAPP_MCDISK': 284,
'IN2P3-LPSC_MCDISK': 143, 'GRIF-LAL_MCDISK': 493,
'RO-02-NIPNE_MCDISK': 184, 'BEIJING-LCG2_MCDISK': 129,
'RO-07-NIPNE_MCDISK': 171}

File access modes used by Ganga Jobs

Two access modes may be used during ST tests.

  • (j.inputdata.type='DQ2_LOCAL') : Currently if users are submitting Ganga jobs with the LCG backend, the posix I/O access is the default method to access files
  • (j.inputdata.type='FILE_STAGER') : One of the alternative access mode used by ST tests is the FileStager mode, which copies the input files in a background thread of the athena event loop usinglcg-cp from the local SE to the worker node tmp area. This mode still needs a bit of improvements to gain full stability, but as demonstrated good performance during ST at some (but not all) sites.