NTuple Production

ntuple_production.jobscript provides a justin jobscript that can run simple larsoft analysis modules on 1 or multiple input files.

Within the justin job, ntuple_prod_utils.py is used to run the lar command and also to create metadata for the output files.

Setup

First, tar up ntuple_prod_utils.py and upload to justin

tar -c ntuple.tar ntuple_prod_utis.py
ntuple_dir=`justin-cvmfs-upload ntuple.tar`

Required environment variables to pass to justin

--env FCL=[fcl_file] (which fcl file to run)

--env NTUPLE_DIR=$ntuple_dir (the cvmfs location of the uploaded ntuple.tar file)

--env DUNE_VERSION=[version] (which version of dunesw to set up)

Optional environment variables to pass to justin

--env NEVENTS=[nevents] (The number of events to run through lar. Default is -1 [all available events])

--env NFILES=[nfiles] (The [max] number of files which will be used requested from justin in each job. Default is 1. Note that justin can submit more jobs than there are input files, so the pool of available files could be exhausted before a job can actually request all files. This is not a problem, it just means that a given output file might be smaller than the rest.)

--env PREFIX=[prefix] (The prefix to the output root file)

--env TREE_NAME=[tree_name] (The tree within the output root file that will be used for some metadata -- TODO might make this more flexible i.e. what if no TTree in output?)