cmssw icon indicating copy to clipboard operation
cmssw copied to clipboard
verified publisher icon cms-sw

A safe guard on GEM DQM harvester, a backport to 12_4_X

Open quark2 opened this issue 3 years ago • 3 comments

PR description:

There is an issue with too large memory consumption in the GEM DQM harvester procedure (#38976). To fix this issue, additional lines are added to suppress the too high number of chambers. This fix is temporary because the origin of the too large number is still missing. This is why I called it a 'safe guard.'

PR validation:

The code of the run issued in #38976 is run again, and I've checked that the memory usage (RSS) is reduced.

If this PR is a backport please specify the original PR and why you need to backport that PR. If this PR will be backported please specify to which release cycle the backport is meant for:

This is a backport of #39060

quark2 avatar Aug 15 '22 10:08 quark2

A new Pull Request was created by @quark2 for CMSSW_12_4_X.

It involves the following packages:

  • DQM/GEM (dqm)

@emanueleusai, @ahmad3213, @cmsbuild, @jfernan2, @pmandrik, @micsucmed, @rvenditti can you please review it and eventually sign? Thanks. @watson-ij this is something you requested to watch as well. @perrotta, @dpiparo, @qliphy, @rappoccio you are the release manager for this.

cms-bot commands are listed here

  • Backported from #39060

cmsbuild avatar Aug 15 '22 10:08 cmsbuild

@cmsbuild, please test

mmusich avatar Aug 15 '22 11:08 mmusich

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-3cb189/26823/summary.html COMMIT: 41f7b588da006f2d47fdfd157b18df40a68edb94 CMSSW: CMSSW_12_4_X_2022-08-15-1100/el8_amd64_gcc10 User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/39061/26823/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

  • No significant changes to the logs found
  • Reco comparison results: 6 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3677242
  • DQMHistoTests: Total failures: 13
  • DQMHistoTests: Total nulls: 1
  • DQMHistoTests: Total successes: 3677206
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 0.004 KiB( 49 files compared)
  • DQMHistoSizes: changed ( 312.0 ): 0.004 KiB MessageLogger/Warnings
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: no differences found

cmsbuild avatar Aug 15 '22 15:08 cmsbuild

urgent

  • needed for data-taking to solve the issue at cmsTalk

mmusich avatar Aug 15 '22 16:08 mmusich

+1

emanueleusai avatar Aug 16 '22 06:08 emanueleusai

This pull request is fully signed and it will be integrated in one of the next CMSSW_12_4_X IBs (tests are also fine) and once validation in the development release cycle CMSSW_12_5_X is complete. This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @qliphy, @rappoccio (and backports should be raised in the release meeting by the corresponding L2)

cmsbuild avatar Aug 16 '22 06:08 cmsbuild

for the record, I profiled the RSS vs time for the job failing at Tier0 (tarball at /afs/cern.ch/user/c/cmst0/public/PausedJobs/Run2022C/HarvestMem/Express_Run357271_StreamExpress) using https://github.com/cms-sw/cmssw/pull/39060 and this is what I see:

image

so indeed there is some sizeable reduction of the peak RSS using this PR.

mmusich avatar Aug 16 '22 10:08 mmusich

backport of https://github.com/cms-sw/cmssw/pull/39060

perrotta avatar Aug 16 '22 16:08 perrotta

+1

  • Temporary solution. The actual fix will likely be implemented in some future cycle

perrotta avatar Aug 16 '22 16:08 perrotta