DQM Offline GUI indexing problem in StreamExpress__Run2022C-PromptCalibProdSiStripHitEff dataset

[rvenditti]

The dataset /StreamExpress/Run2022C-PromptCalibProdSiStripHitEff-Express-v1/ALCAPROMPT (run numbers 355807-355942) is not registered in the Offline DQM GUI. We found that the rootfiles are available on DQM machines (see https://cmsweb.cern.ch/dqm/offline/data/browse/ROOT/OfflineData/Run2022/StreamExpress/0003559xx/ ) but they have not been registered in the DQM Offline GUI database. From log files, we found that the registration failed with following error:

/data/srv/state/dqmgui/offline/data/OfflineData/Run2022/StreamExpress/0003558xx/DQM_V0001_R000355869__StreamExpress__Run2022C-PromptCalibProdSiStripHitEff-Express-v1__ALCAPROMPT.root: error reading file: DQMStore: Monitor element path name 'TEC+1: Map of missing hits' uses unacceptable characters
2022-07-20 01:19:31.380953 [visDQMImportDaemon/2757] imported /data/srv/state/dqmgui/offline/data/OfflineData/Run2022/StreamExpress/0003558xx/DQM_V0001_R000355869__StreamExpress__Run2022C-PromptCalibProdSiStripHitEff-Ex
press-v1__ALCAPROMPT.root with status 256 in 0.566s
2022-07-20 01:19:31.381240 [visDQMImportDaemon/2757] command failed with exit code 256

It has been suggested that this can be due to: https://github.com/cms-sw/cmssw/blob/master/CalibTracker/SiStripHitEfficiency/interface/SiStripHitEfficiencyHelpers.h#L56-L69

@cms-sw/alca-l2 @mmusich @cms-sw/dqm-l2 @pmandrik

[cmsbuild]

A new Issue was created by @rvenditti .

@Dr15Jones, @perrotta, @dpiparo, @rappoccio, @makortel, @smuzaffar, @qliphy can you please review it and eventually sign/assign? Thanks.

cms-bot commands are listed here

[makortel]

assign dqm

[cmsbuild]

New categories assigned: dqm

@jfernan2,@ahmad3213,@micsucmed,@rvenditti,@emanueleusai,@pmandrik you have been requested to review this Pull request/Issue and eventually sign? Thanks

[tvami]

FYI @mmusich

[mmusich]

@rvenditti
thanks for the report, however

It has been suggested that this can be due to:

the problem actually originates elsewhere, actually here:

cmssw/CalibTracker/SiStripHitEfficiency/plugins/SiStripHitEfficiencyWorker.cc

Line 335 in 5984a2e

(isTID ? std::vector<std::string>{"TID-", "TID+"} : std::vector<std::string>{"TEC-", "TEC+"});

Monitor element path name 'TEC+1: Map of missing hits' uses unacceptable characters

I guess the problem is with the '+' sign. Is there a list of characters prohibited somewhere, so that one could be aware of what needs to be sanitized?

[jfernan2]

Hi @mmusich
The problem is in the semicolon ":", not in the sign:
Valid characters:
https://github.com/cms-DQM/dqmgui_prod/blob/af0a388e8f57c60e51111585d298aeeea943367f/src/cpp/DQM/DQMStore.cc#L56

The check:
https://github.com/cms-DQM/dqmgui_prod/blob/af0a388e8f57c60e51111585d298aeeea943367f/src/cpp/DQM/DQMStore.cc#L1893-L1898

[mmusich]

@jfernan2,
thanks

The check:

does it apply only to the histogram names or also to the titles, labels, etc?

[jfernan2]

MEs names and paths, not titles or labels.

DQMStore: Monitor element path name 'TEC+1: Map of missing hits' uses unacceptable characters

[makortel]

Is the validity of the histogram name not checked at the time of booking? If not, could the check be added there?

[mmusich]

Is the validity of the histogram name not checked at the time of booking? If not, could the check be added there?

apparently it's not because running the workflow locally creates valid root files that can be inspected with TBrowser just fine.
The issue appears only in streaming the files on the DQM GUI.

[jfernan2]

Right, it is the index which has troubles with special characters, due to DB constraints

[mmusich]

see a possible fix at #38823. I profited to rename the histograms a bit.

[makortel]

Could a check for valid histogram names be added to histogram booking in order to catch these kinds of problems early?

[mmusich]

Could a check for valid histogram names be added to histogram booking in order to catch these kinds of problems early?

What about here:

cmssw/DQMServices/Core/src/DQMStore.cc

Lines 63 to 68 in 106b43f

	MonitorElement* IBooker::bookME(TString const& name,
	MonitorElementData::Kind kind,
	std::function<TH1*()> makeobject,
	bool forceReplace /* = false */) {
	MonitorElementData::Path path;
	std::string fullpath = cwd_ + std::string(name.View());

[jfernan2]

I leave it up to @emanueleusai to decide, I believe it may be useful, yes

[mmusich]

I believe it may be useful, yes

OK, I've opened #38831 for review. It can be merged in case it's deemed useful by the DQM core team.