Reprocessing: Meeting of 13-Sep-2004: 9:30-10:30 ESNet video conference

Agenda

1. News
2. Status and Issues in Reprocessing through JIM (Gabriele Garzoglio)
3. AOB.

Minutes

Participants

Topics

1. News:
   
   • First release of p17 is frozen. Ian needs to create the RTE tar-ball, which needs some more work than usual due to the switch to gcc.
     Before remote site jump into testing, one round of testing at Fermilab is foreseen.
     
   • Calorimeter phi calibration constants not ready until mid-November.
   • Data distribution
     Mike suggests to reuse original processing projects for downloading. UTA is in contact with Mike.
     
   • Network
     Westgrid should have 1Gbit/s to Fermilab (10Gbit/s soon) though for testing only.
   • CERN Courier is writing about our p14 reprocessing
2. Status and Issues in Reprocessing through JIM (Gabriele Garzoglio)
   • Ian: Request system: Reprocessing requests might fit into the current system.
     
   • Several version of d0reco, recocert, recotests might be needed
     
   • Model: Size of grid jobs should be of the order of 100 files per job. One run per GridJob feasable? Larger number more difficult to recover in case of error.
   Further Discussion:
   • Plan: GG gets datasets working (estimate is that it takes a couple of weeks). When it's working Farms should try it out. This will require a JIM update.
   • TN: How do site specific paramters, rcp parameters get entered into the job?
     GG: Local configuration of JIM passed to worker nodes by JIM. Calibration DB name needs to be specified locally. IB: Added to runjob, site configurable.
   • GG: Which parameters are site specific?
     We discusssed alread: DB name, number of files per job. (Are there further ?)
   • TN: How to recover from crashes?
     Outside JIM, resubmit with new dataset? Possibilities on the spot inside runjob?
   • PL, TK: What shall happen to the DST output?
     DSTs will be abandoned and most likely not stored.
3. AOB:
   -