Workflow Team Meeting - June 11 4PM CERN, 9 FNAL time

Vidyo Link

Attending

  • FNAL: SeangChan will not be able to make the meeting, Jen & Luis, Jorge
  • US: Ajit, Matteo, Stephan
  • CERN : Julian, Andrew and Alan, JeanRoc, Dima
  • EU:

Personel

  • Jen off June 27-July5 (tenitive)
  • Jen off Aug 10-26 (tenitive )
  • SeangChan July 27-31

News

  • New Luis and New John have accepted our offers, we are working on getting their visa's and hope to get them to Fermilab in July so we can have overlap in these positions.
  • there may be a re-running of miniaod, high IO but low cpu time, 1.5 billion events
    • Dima: I don't know where this information is coming from. PPD has no plans to do miniAOD on MC in near future (we just produced that). Anyway, the only real issue that we may have with miniAOD campaigns is to have a proper setup to handle it, which is not very different from DigiReco. I/O load should be much smaller for miniAOD since we don't need to read hundreds of minbias events per one collision event as we do for DigiReco.
  • We are back to needing US operators, with data taking starting again, we need to look over what we are doing and possibly re-define roles
    • Dima: we are trying to shift some responsibilities to the central computing shifters. Christoph Wissing needs to test the instructions and integrate them.
    • Jen will reach out to Sean and see if he can help us over the summer
  • on scale of 1 mo we will have DigiReco

3 top issues effecting production

  • unable to clone or make ACDC's on Tues
  • Robust merge issue - Alan was supposed to look at it
  • DBS3 open blocks and recovery script for loss of miniaod

Site support - John

Waiting Room

Workflows

ReDigi

TaskChains

  • (JR) what can we do to mitigate the taskchain job multiplication (1 job first task, 500 jobs second task) :
    • blow up splitting of first task to cover more sites ?
    • have the second step relocalised ?
    • we need to put them at sites that can handle the large 2nd stepsl
      • lands someplace for first task, and runs it then second step spawns 500 jobs that need to read that output.
      • we need to carefully assign these TaskChains to sites that are larger. The 2nd step needs to run at the same site as the first step ran, output is not in dbs and PhEDEx yet so we can not read by xrootd

Rereco

Store Results

MonteCarlo

Agent Issues

Redeployment Plan

  • nope - started validating new release yesterday

RelVal Andrew

  • SeangChan will not be able to attend the meeting. Andrew should ask Alan all questions.
  • he will wait until next week

L3 discussion - Ajit, Jean-Roch, Matteo

  • things are slow so not much to discuss

Opportunistic Resources - Stefan

  • infrastructure upgrades so not much going on. Furruk sent test jobs, they eventually ran

HLT

SDSC

Automatic Assignment And Unified Software

AOB

-- JenniferAdelmanMcCarthy - 2015-06-10

Edit | Attach | Watch | Print version | History: r4 < r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r4 - 2015-06-11 - JenniferAdelmanMcCarthy
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback