LEM processing using the dropbox¶
Some studies require the PIDs to be rerun. It is not possible to run LEM onsite at Fermilab, the grid nodes have insufficient memory. Instead you can use the LEM dropbox. Simply place your files in a magic directory, automated scripts will process them at Caltech, and the results be returned to you. It should be completely fire-and-forget
Make sure you actually need to rerun LEM¶
You only need to rerun the PID step if you've changed something in the simulation or reconstruction steps over the standard files. If your files are standard they've probably already been through the official processing. Please look to see if what you want is here https://cdcvs.fnal.gov/redmine/projects/nova_sam/wiki/Official_Datasets
Creating dropbox directories¶
mkdir /nova/ana/nu_e_ana/lem_dropbox/$USER/ chmod g+wx /nova/ana/nu_e_ana/lem_dropbox/$USER/
It's important to set the permissions so that the processing scripts can create and remove the files and subdirectories they need.
You probably want to make a subdirectory so that you can keep track of different processings
mkdir /nova/ana/nu_e_ana/lem_dropbox/$USER/mysubdir/ chmod g+wx /nova/ana/nu_e_ana/lem_dropbox/$USER/mysubdir/
Putting files in the dropbox¶
Then symlink the files you want to be LEM'd (ie .reco.root or .pid.root files) into mysubdir/
They'll immediately start disappearing, to be gradually replaced sometime later (O(1 day)) with .reco.lem.root or .pid.lem.root files. You'll notice that the output files are actually symlinks to pnfs locations, and in fact that's the recommended thing to link in in the first place. Trying to do this processing on bluearc tends to waste a lot of time waiting for CPN locks.
ln -s /pnfs/nova/scratch/users/`whoami`/<myfiles>/*.pid.root /nova/ana/nu_e_ana/lem_dropbox/$USER/mysubdir/
Please let me (email@example.com) know when you do this so I can keep an eye on things. The process is 100% automated, but I have to start a few scripts up (I haven't been able to get
kcron to work 100% reliably) and other things do sometimes go wrong.
You can see progress yourself at http://nusoft.fnal.gov/nova/lem/ but I have access to more of the pieces to figure out why things might have stalled
If your files need special handling in some way (unusual module labels, special settings required during processing) please contact me. The ability to handle many of these special cases is built in, but not worth going into here.
LEM'ing. Oh, wait...