Description of bug
Ticket: ATLMCPROD-8469 DSIDs: 500794, 500795 There was a crash in official production at the very end of the madspin_on_lhe() function, which performs MadSpin decays on input lhe files from MadGraph. This crash occurs if you pass a tarred file to to madspin_on_lhe() (e.g. TXT.440006._000388.tar.gz) but for an untarred file (e.g. TXT.440005._000822.events). Error message: 12:13:06 Py:MadGraphUtils INFO Unzipping generated events. 12:13:07 gzip: TXT.440006._000485.tar.gz.gz: No such file or directory
Changes introduced
Ugly workaround, untarring the inputGeneratorFile and passing that to madspin_on_lhe(). It is what I got to work. I see a pipeline failing because runArgs.inputGeneratorFile is specified in the JOs. Would the jobs crash in official production (they ran locally) or is this just bad practice?
Tests
Tested in current mode and in a mode where the output file from MadSpin was removed (to make sure it is actually used, which it is).
Issues resolved
In my local test, yes. Closes #