Skip to content

[v7r1] Improve backwards compatibility for running jobs submitted with v7r2#5177

Merged
atsareg merged 1 commit into
DIRACGrid:rel-v7r1from
chrisburr:back-compat
Jun 13, 2021
Merged

[v7r1] Improve backwards compatibility for running jobs submitted with v7r2#5177
atsareg merged 1 commit into
DIRACGrid:rel-v7r1from
chrisburr:back-compat

Conversation

@chrisburr

Copy link
Copy Markdown
Member

As part of the upgrade to v7r2 in LHCb we found jobs started crashing with:

2021-06-03 06:53:27 UTC dirac-jobexec/GaudiApplication ERROR: Failure in GaudiApplication execute module
Traceback (most recent call last):
  File "/cvmfs/lhcb.cern.ch/lhcbdirac/v10r1p13/LHCbDIRAC/Workflow/Modules/GaudiApplication.py", line 164, in execute
    prodConfFileName = self.createProdConfFile(stepOutputTypes, histogram, runNumberGauss, firstEventNumberGauss)
  File "/cvmfs/lhcb.cern.ch/lhcbdirac/v10r1p13/LHCbDIRAC/Workflow/Modules/ModuleBase.py", line 1108, in createProdConfFile
    jobMaxCPUTime=86400)
  File "/cvmfs/lhcb.cern.ch/lhcbdirac/v10r1p13/LHCbDIRAC/Workflow/Modules/ModulesUtilities.py", line 132, in getEventsToProduce
    CPUTime = getCPUTime(CPUNormalizationFactor)
  File "/cvmfs/lhcb.cern.ch/lhcbdirac/v10r1p13/DIRAC/WorkloadManagementSystem/Client/CPUNormalization.py", line 219, in getCPUTime
    queueSection = '/Resources/Sites/%s/%s/CEs/%s/Queues' % (siteName.split('.')[0], siteName, gridCE)
AttributeError: 'NoneType' object has no attribute 'split'

The problem is that the LocalSite part of the CS not being filled as the pilot.cfg file is missing.

The cause is that the Executable in the job wrapper is now dirac-jobexec instead of $DIRACROOT/scripts/dirac-jobexec. DIRAC v7r1 only copies the pilot.cfg if $DIRACROOT is found in the Executable name so existing pilots running v7r1 with filling mode fail to execute jobs that were submitted with v7r2.

Nothing can be done to fix the pilots in LHCb but this should make upgrading smoother for other installations.

BEGINRELEASENOTES

*WorkloadManagement
FIX: Look for dirac-jobexec instead of $DIRACROOT to decide when to copy the pilot.cfg to improve compatibility with v7r2

ENDRELEASENOTES

@chrisburr
chrisburr requested review from atsareg and fstagni as code owners June 3, 2021 09:34
@atsareg
atsareg merged commit cb6cad0 into DIRACGrid:rel-v7r1 Jun 13, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants