Fix all DQM paths to properly handle TriggerResults #15810
Conversation
Instead of asking for the default TriggerResults and assuming it is the HLT one, explicitly ask for the HLT results.
When running with fastsim, these modules need to ignore the HLT TriggerResults. Previously this was done by moving the Sequence they were on from an EndPath to the Path. But with an upcoming framework change this will no longer be allowed. Instead we now explicitly turn off that functionality when running under fastsim.
Fix a segmentation fault which can occur if the miniAOD filters are checked when not diong miniAOD MET. This fix allows METAnalyzer to be properly used on EndPaths.
Many of the modules on the different dqmoffline*_step paths require access to the TriggerResults. Previously those paths were sometimes made Paths and other times made EndPaths in an attempt to only pick up the TriggerResults that were actually needed. Now all the modules have been explicitly changed to ask for the TriggerResults using the process name so it is always safe to use an EndPath. In addition, an upcoming framework change will enforce that all data requested must be accessible to a module. Therefore all modules asking for the TriggerResults without specifying a process name (or specifying the present process name) must be on EndPaths.
|
A new Pull Request was created by @Dr15Jones (Chris Jones) for CMSSW_8_1_X. It involves the following packages: Configuration/Applications @cmsbuild, @dmitrijus, @franzoni, @vanbesien, @davidlange6 can you please review it and eventually sign? Thanks. cms-bot commands are list here #13028 |
|
please test |
|
The tests are being triggered in jenkins. |
|
-1 Tested at: 965e6d1 You can see the results of the tests here: I found follow errors while testing this PR Failed tests: RelVals
When I ran the RelVals I found an error in the following worklfows: runTheMatrix-results/140.53_RunHI2011+RunHI2011+RECOHID11+HARVESTDHI/step2_RunHI2011+RunHI2011+RECOHID11+HARVESTDHI.log |
With dqmoffline_step now always being an EndPath, we need to be sure that modifications to Paths are also applied to EndPaths.
|
Pull request #15810 was updated. @cmsbuild, @dmitrijus, @franzoni, @vanbesien, @davidlange6 can you please check and sign again. |
|
please test |
|
The tests are being triggered in jenkins. |
|
Comparison is ready @slava77 comparisons for the following workflows were not done due to missing matrix map:
|
|
hold |
|
Pull request has been put on hold by @Dr15Jones |
|
unhold |
|
Ran the full runTheMatrix.py and all workflows passed. |
|
This fix is needed to allow further work on the threaded framework. |
|
ping |
1 similar comment
|
ping |
|
ping |
|
+1 |
| else: | ||
| # schedule DQM as a standard Path | ||
| setattr(self.process,pathName, cms.Path( getattr(self.process, sequence) ) ) | ||
| self.schedule.append(getattr(self.process,pathName)) |
There was a problem hiding this comment.
I suspect this removal caused loss of MiniAOD validation histograms in 810pre12, see
https://hypernews.cern.ch/HyperNews/CMS/get/relval/6213/46/2.html
I have a proposal for a fix in #16265.
This pull request fixes all problems from issue #15769 and includes the changes in #15809. The most important additional change on top #15809 changes ConfigBuilder.py and therefore could have a larger impact on the evaluation of this pull request as opposed to #15809.
The ConfigBuilder.py change always makes the dqmoffline*_step path be an EndPath. Previously, based on which processing step is being configured the dqmoffline*_step might sometimes be a Path and other times be an EndPath. The reason appeared to be because sometimes the TriggerResults was being requested by some modules using an empty processName and the hope was one would pick up the HLT results. This change has all modules explicitly state what process name to use when getting the TriggerResults.
This is needed since the next threaded framework change requires that the consumes of a module and its place on a Path or EndPath guarantee that it can get all the data it requests (or will fail since there is no source or module producing the data it wants). Asking for a TriggerResults for the present process from a Path is therefore not allowed (and was not a good idea before either).