Skip to content

OSG Technology Area Meeting, 15 June 2020

Coordinates: Conference: +1 312-626-6799, PIN: 718 161 330, https://cern.zoom.us/j/718161330 (password sent separately)
Attending: Mat, BrianL, Diego, TimT, Carl, Edgar, Marian, Marco Mambelli

Announcements

  • Brian giving HTCondor-CE webinar at EGI; reached out to team members for help during Q&A.
  • Doc focus this Thursday

Triage Duty

  • This week: Edgar
  • Next week: Mat
  • 9 (+2) open FreshDesk tickets
  • 0 (+0) open GGUS ticket

JIRA

# of tickets Δ State
150 +2 Open
42 -1 In Progress
10 -4 Ready for Testing
0 +0 Ready for Release

OSG Software Team

  • OSG 3.5.18
    • AI (BrianL): Release HTCondor-CE 4.3.0+
    • AI (Diego): Build XRootD 4.12.2 (SOFTWARE-4063)
    • AI (Diego): Build XRootD plugin .so's based on XRootD version (SOFTWARE-4093)
      • Waiting on review from Derek on HDFS plugin.
  • GSI/GridFTP migration
    • AI (Edgar): test new GFAL client available in EPEL testing with XRootD 4 and 5
      • There does not seem to be a new GFAL client in EPEL testing for EL 7
    • AI (Edgar): Update a VO frontend to GlideinWMS 3.7
    • AI (Carl): OASIS manager + COManage endpoint (SOFTWARE-3947)
  • Enterprise Linux 8
    • AI (Edgar): Test built XRootD packages for EL8 (lower prio)
    • AI (Mat): Build osg-wn-client tarball (SOFTWARE-4050)
    • AI (Mat): Get CentOS 8 images ready for VMU tests (SOFTWARE-4072)

Discussion

  • CentOS repos were broken Friday/weekend, causing false errors in VMU tests.

  • BrianL sent team members filtered spreadsheets for the SW WBS; please prioritize tasks marked OVERDUE.

  • GlideinWMS

  • Marco found problem that when a grid universe job lands against an invalid grid resource, it remains idle forever, instead of going on hold. Try reproducing on 8.8 and send email to condor-users.
  • Denis still working on token auth between frontend and collector; might be a good idea to do some pair programming with an OSG developer so we can help with HTCondor config etc.

Support Update

  • BrianL: helped AGLT2 with problem where home dir was on a shared file system and condor was incorrectly assuming the shared file system was mounted on the execute host as well as the submit host, and not doing file transfer.

  • Edgar: helped LIGO with interpreting graphs. We have some bad names in there like how "batch records" are actually "pilot records". Also the meaning of "dedicated" vs "opportunistic" isn't obvious. Also Syracuse is listed as owned by LIGO which is wrong. Edgar should talk to Derek about making some of these easier to understand.

  • Marco: GlideinWMS 3.7: using TOKEN auth results in tarballs that have some libraries from the OS. This should be fixed in 3.7.1.

OSG Release Team

3.4.53 Δ Both Δ 3.5.19 Δ Total Δ Status
0 +0 0 -1 5 +0 5 -1 Open
0 +0 0 +0 10 +2 10 +2 In Progress
0 +0 0 -2 4 -4 4 -6 Ready for Testing
0 +0 0 +0 0 +0 0 +0 Ready for Release
0 +0 0 -3 19 -2 19 -5 Total
  • Software
    • Ready for Testing
      • 3.5.19
        • scitokens 0.7
        • Upcoming: HTCondor 8.9.7
        • xrootd-lcmaps 1.7.7
        • Upcoming: XRootD 5.0.0
    • Ready for Release
      • Nothing yet
  • Data
    • Nothing
  • Operations
    • Nothing
  • Contrib
    • Nothing

Discussion

  • HTCondor 8.9.7 will be re-released this week but we need to review Jason's update instructions to publish them ourselves and/or modify the packaging so that manual interventions aren't required

OSG Investigations Team

  • Derek was on slate training.
  • GRACC 2 switch was performed today. Lots of investigation effort to figure out bottlenecks.
  • Lots of communication with XRootD team on TLS + Tokens support.

Discussion

  • Edgar's student starts in July to do scale testing for HTTP third party copy. Considering testing authenticated CVMFS but might not be worth doing it until the SciTokens switch (since we'd just have to do it again).