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

Agenda

1. News
2. Status of Implementation of Reprocessing in JIM (update)
3. Status of JIM Deployment to Remote Site
4. AOB.

Minutes

Participants

Topics

1. News
   p17.02.00 failed to process post shutdown data (problem with calibration database).
   
2. Status of Implementation of Reprocessing in JIM (update)
   
   • Processing
     
     • New parentage:
       
       • Question of necessity by Heidi.
         Andrew uses it to track file relations, i.e. to check whether a file is already merged. Without this the test becomes more complicated.
       • Question of speed for analysers?
         We can fill a direkt link? There is a field available in v5.
         One would have to check which database server actually support this. Modifications needed in runjob.
         Decision: in case its needed for performance reasons we modify the metadata after the fact.
     • Systematic way of storing runjob tar-balls: (Wyatt, Gabriele and Iain should coordinate)
       Short term: naming convention.
       Medium term: use metadata. All tar-balls should be stored to enstore (by Iain). Names need to be agreed with the JIM team still.
       Main directory mcc-dist for MC, RTE_runjob for RTE tar-balls (to be used for reprocessing) is not going to change for mc_runjob related tar-balls. Iain didn't want to promise for tar-balls made for use with d0_runjob.
   • Merging
     How do we mark tests in the metadata? Mark test outputs as bad.
   • Production scripts
     
3. Status of JIM Deployment to Remote Site
   
   • DØFarm: Farm was upgraded to Scientific Linux. Working on testing p17.01.00.
     
   • GridKa: Out of business for reconfigurations
     
   • Lyon: p14 processing and merging worked. Proxy database is actually used.
     
   • SAR: Jae by email: large test of production succeeded to store to durable location.
     
   • WestGrid:100 job test. Run through, but some had problems with storing (2 of 99 files affected). I.e. ready for migrating to new sam station. Uses remote production router
     
   • Wisconsin: no update
     
   • CMS Farm: no update
     
   • UK: Imperial: awaiting email by Frederic
     
   • Prague: Recent versions installed. Problem is XML DB went out of file descriptors. Manual cleanup needed for farms which run in the old scheme for a long time.
     
4. AOB:
   

   Iain: MC Farms have problems with storing files the produce. This might be a bottleneck for reprocessing. Might be due to the recent SAM-Nameservice problem.

   Laurent: Scientific Liunx was tested at IN2P3. Zee recocert output was OK.

Next Meeting