05102013 Minutes


Himansu, Ji, Minerba, Jarek, Dominick, Susan, Nick, Kanika, Jan, Chris, Nate, Robert, Vladimir, Gavin, Ryan, Raphael, Adam

LEM off-site production (Chris)

Output of LEM is essentially just PID numbers. Input is just a list of cell-PE number pairs (LiteHits).
One stream of file is regular data file and the second is a summary file. Summary file is ~1MB from ~several GB. This we can then transfer to Caltech (the only place we can process the current incarnation of LEM) in a timely fashion. The upload rate is ~1MB/s so summary files are necessary because without them it would take over 200 DAYS to transfer ~20TB of MC/Data.

SAM Metadata progress (Adam)

Metadata model is progressing well. Can now input metadata into a file via a fcl job. Can also parse filenames to put into the metadata. Spoke with Robert I and Marc M about uploading files to SAM database. Robert I needs to know what the files now look like with the metadata

CVMFS (Raphael)

Successfully could use glide-in to Fermilab from Harvard. Vladimir tried to glide-in from SMU using cvmfs and mostly successful, except failures on nodes without CVMFS. Showing real promise for off-site production in the next phase.

Craig Group informs us that FermiGRID is down next day on Thursday May 16th. ALL GRID jobs will be terminated and not restarted during the downtime.

Software Update (Jan)

Utilities package rewritten to have a func/ non-ART dependent package. Denver discovered that his package that used LinFit could not find the right library. Just need to add and all is well with the world, no major rewrite elsewhere.

SVN repository for all has been created by Jan called devs. Jan is working on putting in the hooks to stop upload of large files! Punishment is a Production task ;)

Tagged S13-05-09 for running LEM tests. Jan will backport the Utilities fix to the tag, but only re-build after LEM tests have concluded.

Processing FarDet and NDOS data (Jarek)

Biggest issue is that somebody needs to be assigned to this task!