(Almost) empty kappa files

[thomas-mueller]

In the latests Higgs skim, I see several files without entries in the Events tree and one entry in the Lumis tree:

get_entries.py dcap://dcache-cms-dcap.desy.de//pnfs/desy.de/cms/tier2/store/user/fcolombo/higgs-kit/skimming/2015-10-14/SingleElectron/crab_SingleElectron_Run2015D_PromptRecov4_13TeV_MINIAOD/151019_073217/0000/kappa_SingleElectron_Run2015D_PromptRecov4_13TeV_MINIAOD_240.root Lumis: 1 entries Events: 0 entries

This gives problems when they are processed by Artus:

Lumi section 1850307694:1702249803 not found or unique! 0x3ba3940 0 dcap://dcache-cms-dcap.desy.de//pnfs/desy.de/cms/tier2/store/user/fcolombo/higgs-kit/skimming/2015-10-14/SingleElectron/crab_SingleElectron_Run2015D_PromptRecov4_13TeV_MINIAOD/151019_073217/0000/kappa_SingleElectron_Run2015D_PromptRecov4_13TeV_MINIAOD_240.root {(258692, 1): 0}

I can avoid this error, when I run over more than one file (which we usually do not do in case Svfit needs the be run).

Has anybody else observed this problem? Was there probably a change in the filling of the Lumis tree or even in the miniAODv2, because we did not see that in the miniAODv1 skims? Anyway, I think, it is easy to fix in the FileInterface in KappaTools. One could just check if there are entries at all in the Events tree and otherwise ignore this check.

[thomas-mueller]

Fixed by https://github.com/KappaAnalysis/KappaTools/commit/8f9039f9b7724852c1be352a3f220f6c563c2937 and https://github.com/artus-analysis/Artus/commit/742d39a024b7224ddfd5594b4221277ba574af1c. You need to pull in KappaTools and compile with make. You might need further fixes like https://github.com/artus-analysis/Artus/commit/2847376d785c055e5e4bffe0033cb05fdd3bd49b to prevent consumers of saving objects that are not initialised if no events are processed.

[thomas-mueller]

The highly sophistcated Makefile structure requires to compile it with make -B. This way you also profit from some minutes of rest. ;-)