Tier1 and Tier-2 Weekly Report 2017-02-14

General News

  • ADC Tech Coord Meeting [20170213]
  • ADC Weekly Meeting [20170214]
    • MC Production
      • Down to about 290M events in production queue
      • Present fairshare for MC prod is 75%
        • Two global fairshares: MC16 ~1%, MC default ~ 73%
        • Will push these to 50/50 after monitoring
      • Start MC16 sim tasks asap. Optimized for a month's production ~ 500M events
      • Expect to start MC16 digi+reco end of the month
    • Group Derivation Production
      • Both data15+16 nearly finished, apart from few stragglers
      • debug streams also complete
      • Build 20.7.8.12 cache yesterday
      • Still working on DAOD size reduction survey
    • Reprocessing
      • Planning being discussed - want about 75k slots for ~2 months for 2016 data and ~1 month for 2015 data. Been delayed by 6+ months.
    • CRC/ADCoS Report
      • Got up to 300k slots with CERN P1 (50k cores) and Titan
      • Some tape tests going on with T1's last week
        • Stage 150TB from each T1 tape (300TB at BNL)
        • Ran into issue with timeouts set in FTS (48 hours) and defaults on sites
        • Full report next week
      • DDM functional tests have not been running since 18 Jan due to failure to upload source files - not sure this is necessary?
      • FTS race condition resurfaced leading to lost files
      • Overflow jobs have been stopped because new pilot mover requires data to be local
      • Jobs requiring 1000 tiny EVNT files:
        • Tasks are forced to nucleus to avoid data transfer
        • Still causes problems in NDGF since ARC CE has to copy them
        • EVNT merging is currently under test
      • CERN P1 and SuperMUCboth running event service stressed CERN S3 object store - turned out there was one bad node causing problems
      • Live page updates for shifters been improved and CC7 compatible
      • No AFS day TOMORROW
        • Reminder: CERN AFS will be blocked outside CERN for 24h starting 9am tomorrow
        • Some SW releases are known to try and access libraries on afsfirst before cvmfs
        • If CERN afs is mounted it is likely those jobs will hang
        • Recommendation: unmount CERN AFS if you can
  • CA Cloud Squad Actions
  • Open Issues and Ongoing Discussions
  • WLCG and General Computing News
    • Security News
    • gLite/EMI News
  • ROC_Canada News

ABCD Determination and Monitoring

SSB Results

SAM Results (based on VO tests)

APEL Accounting

APEL Synchronization Test Links

Production and Usage Status

-- %USERSIG{LeslieGroer - 2017-02-14%

Comments

Topic revision: r1 - 2017-02-14 - LeslieGroer
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2026 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback