CAF instructions for making TopNtuples

You must work in fcdflnx3, because that's where the TopFind executable was linked. Follow these steps:

• fcdflnx3> setup cdfsoft2 development
  • don't worry this is not 4.11.1, we need development for command line CafSubmit to work
  • version 4.11.1 is actually setup inside the shell script one sends to the CAF
• fcdflnx3> mkdir myCAF (or any name you want)
• fcdflnx3> cd myCAF
• cp /cdf/home/gervasio/CAF_TopJP4111/.cafrc . (or copy your own .cafrc)
• cp /cdf/home/gervasio/CAF_TopJP4111/TOP_MC.sh .
  • Edit TOP_MC.sh and change "gervasio" for USERNAME in the line:

    fcp ./TopNtuple_${OUTNAME}.root gervasio@${CAFOUT_NODE}:${CAFOUT_SCRATCHPATH}

• cp /cdf/home/gervasio/CAF_TopJP4111/TOP_MC.tcl .
• cp /cdf/home/gervasio/CAF_TopJP4111/CafSubmit_MC.sh .
  • This script send the job TOP_MC.sh to the CAF
  • TOP_MC.sh exports the input and output filenames which are read in by TOP_MC.tcl, so in principle you don't need to change TOP_MC.tcl. You need only change TOP_MC.sh to submit different filesets.
  • You might want to change the "GGG" tar filename to anything you like
• ln -sf /cdf/scratch/gervasio/TopAnaRel_4111/bin bin
• ln -sf /cdf/home/gervasio/TopAnaRel_4111/shlib shlib
• ln -sf /cdf/home/USERNAME/rcpdb personal_rcp
  • USERNAME means sluca, palencia or whatever your user name is
  • If you do not have a rcpdb directory in your home area, you need to create one. Go to your home directory and execute the command "rcp_init_personal"
• Find out which are the datasets which you want to TopNtuplize
• Send one CAF job for each fileset in the dataset
  • Edit TOP_MC.sh
  • Give the output filename in TOP_MC.sh as dataset_fileset (see the TopNtuples done so far)
  • Give the input filename in TOP_MC.sh as the fileset you need
  • You might also have to change the input book name
• You are now ready to submit the job: ./CafSumbit_MC user queue outfile
  • ./CafSubmit_MC with no arguments gives usage and an example
• After job is done (you get an email), make a directory in TOP called as the dataset, and copy all the TopNtuple_dataset_fileset.root there
  • We might have some disk space these days....