Project

General

Profile

JobSub Users Meeting: May 14 2014

Meeting Notes

Present: Parag Mhashilkar, Dennis Box, Eric Flumerfelt, Art Kreymer, Gerard Bernabeu Altayo, Joe Boyd, Yun-Tse Tsai, Neha Sharma, Tanya Levshina

Nova (Eric)

  • AWS: Issues with identifying if files came back.
  • Its hard to figure out where the issues are? In aws? jobsub? user exe?
  • Fetchlog is inconvenient to use
  • Have couple of submission scripts
  • Changes to the jobsub command interface should be communicated in advance to give the users time to adjust to changes

Microboone (Yun Tse)

  • Have different wrappers for submitting using jobsub-tools to Fermigrid and using new jobsub client to OSG
  • Tanya: Do not use CONDOR_TMP area
  • Monte Carlo challenge in July. Until then
    • Plan to have a submission client in SLAC
    • Use jobsub client, single submission script for submitting to both Fermigrid and OSG

LBNE (Tanya)

  • Build larsoft with X11 libraries and shipped everywhere. Larsoft needs x11 libraries which are not present on all the OSG sites. Outside the scope of jobsub.

Minos (Art)

  • Minos uses few wrapper scripts
  • Minos reached upto 6000 jobs running at once since they moved to fifebatch1
  • In general working well.
  • Issues/Suggestions
    • Check the cert and renew it. Uniq to cert & for user
    • Jobsub to use its own proxy on the client
    • Need a single jobsub wrapper to run condor commands
    • Replace condor's jobid with jobsub's jobid