Read nEventsPerJob from JO instead of logfile
Currently the logParser checks whether nEventsPerJob
is set correctly to fulfil the runtime limits of the grid job. It uses the nEventsPerJob
written out to the log.generate
, but that might not be the correct one in cases where nEventsPerJob
had to be changed after the test job. Instead the value in the JO should be used.
These checks should probably be removed from the logParser and moved into an individual check of the JO (together with the log.generate.short
?).
I'm attaching one log.generate file for a recent merge request (DSID 700008) which without modifications leads to:
- CPU = 0.88 hrs <-- ERROR: Too low CPU time - should be between 6-12h. Adjust nEventsPerJob!
Edited by Frank Siegert