Skip to content

OSG Technology Area Meeting, 26 August 2019

Coordinates: Conference: 719-284-5267, PIN: 57363;
Attending: BrianL, Derek, Diego, Mat


  • Next week's meeting (9/2) canceled for Labor Day
  • IRIS-HEP retreat September 12, 13: block your calendars
  • Carl OOO, returning 8/29
  • Edgar OOO, returning 9/3

Triage Duty

  • This week: BrianL
  • Next week: Carl
  • 8 (+3) open FreshDesk tickets
  • 2 (+0) open GGUS ticket


# of tickets Δ State
153 -6 Open
42 -3 In Progress
20 +7 Ready for Testing
2 +1 Ready for Release

OSG Software Team

  • OSG 3.4.34
    • Diego found issues with buiding stashcache-client
    • AI (BrianL): Release XCache 1.1, build, and test
  • OSG 3.5
    • Package list:
    • Perform packaging work in the osg-3.5 and devops SVN branches
    • 3.5.0 will ship without HTCondor-CE or GlideinWMS: new major versions will be targeted for OSG 3.5.1+
    • AI (Mat): Build HTCondor 8.8 with default DAEMON_LIST and pool password (SOFTWARE-3795);
    • AI (Diego): Test osg-configure 3.0.0
    • AI (BrianL): Investigate OSG 3.5 test failures (Slurm configuration)
    • AI (BrianL): Document manual 3.5 upgrade instructions (SOFTWARE-3799)


Diego will update the stashcache-client spec to use upstream's

Support Update

  • INFN (Mat): Assisting with StashCache installation issues (mismatched FQAN, missing dependencies)
  • Milwaukee (Derek): Moving LIGO containers to the OSG CVMFS repo
  • Oklahoma (BrianL): Issues with logrotate and XRootD on hosts that share the same configuration/OS
  • Syracuse (Mat, Edgar): larger GLOW test jobs succeeded but users from the OSG VO ran into similar problems
  • GGUS ticket w/ Doug Johnson about using Let's Encrypt with LVS; still need to create the ticket for load-balanced GridFTP + Let's Encrypt document

OSG Investigations Team

  • CMS requested some changes to xrootd monitoring. Spurred many other changes which are just finishing up.
  • Working with operations on new, better monitoring of CVMFS exporters. LIGO failed undetected for a while. Now fixed, but need better alerting on what ended up being a "WARNING" that should have been a failure.


  • Working with operations investigating OAuth with co-manage. Waiting on co-manage (Wisconsin?!?!)




None this week