Skip to content

OSG Technology Area Meeting, 28 January 2019

Coordinates: Conference: 719-284-5267, PIN: 57363;
Attending: BrianL, Carl, Edgar, Marco Mambelli, Mat, TimT


Operations F2F meeting to Tues-Thurs (

Triage Duty

  • This week: TimT
  • Next week: Carl
  • 6 (+1) open tickets


# of tickets Δ State
146 -6 Open
35 +5 In Progress
4 +0 Ready for Testing
1 -4 Ready for Release

OSG Software Team

  • 3.4.24
    • XRootD 4.9.0 RC4 soon
    • AI (Mat): What's left for XCache 1.0.1?
    • AI (Mat): Update gsi-openssh version
  • XCache container due at the end of the month


  • Hold off on tagging XCache 1.0.1 since container work is still ongoing and finding issues
  • AI (BrianL): follow-up with Marco regarding signal trapping in the blahp
  • Stash Cache and Origin registrations need to authZ updates before the release of XCache 1.0

Support Update

  • AGLT2 (BrianL): high load on some worker nodes (ls /tmp hangs) with some ATLAS jobs using 100% of swap
  • FNAL (BrianL): Issues with the HTCondor accounting fix, Farrukh is in direct contact with the HTCondor team
  • Georgia Tech (Edgar): Initial cache registration is complete
  • Ice Cube (Edgar): Edgar set up a public squid for Ice Cube so that UW-Madison's CMS T2 can retire their public squids
  • Purdue (Marco): Glideins ignoring SIGTERM because the wrapper script created by the BLAHP is trapping signals and not forwarding them
  • UERJ (Mat): Job router is ignoring jobs in the CE queue, discussing the issue with the HTCondor team

OSG Release Team

3.4.24 Δ Status
7 +7 Open
15 +15 In Progress
4 +4 Ready for Testing
1 +1 Ready for Release
27 +27 Total
  • OSG 3.4.24
    • Ready for Testing
      • xrootd-lcmaps 1.6.0
      • Upcoming
        • Singularity 3.0.3
    • Ready for Release
      • Upcoming
        • HDFS-FUSE for EL6
  • Data
    • Nothing
  • Operations
    • Nothing
  • Contrib
    • Nothing


  • AI (TimT): Ping Marian to test xrootd-lcmaps

OSG Investigations Team

Edgar is working with a student to do load tests of authorized reads of LIGO data from a cache. They've noticed some issues when there are thousands of clients trying to read the data and will investigate it further this week.



None this week