AFS weekly meeting

→ Europe/Zurich
31/1-012 (CERN)

31/1-012

CERN

6
Show room on map
Description
AFS service/operation meeting
Participants
  • Dan van der Ster
  • Jakub Moscicki
  • Jan Iven
  • Massimo Lamanna

AFS operations meeting 2016-02-29

present: Dan, Kuba, Jan

Issues:

  • all AFS servers rebooted (glibc update + new kernel) last Wed+Thu
    • 1 unclean shutdown on "user" server = salvage at restart
    • 1 BOINCed(?) CMS server afterwards (but restarted cleanly)
    • Thread limits increased (overall 256->1024, throttling 10/20 -> 40/80). Expect less "overload" mails
    • bosserver.sh script gone
    • 4 mis-assigned spare disks reported after boot - "mdadm --monitor" seems to ignore these? or only warns once?
  • ABS: many errors for last Thu (CASTOR "afsbackup" pool running with 1 diskserver?) - will get cleaned
  • new LEMON metric for offline volumes (seen after restart, user complained)
  • Decision: turn off the Thu morning client stat collection script since it causes instabilities. Run perhaps once/month?
  • Dan: UID-updating script had missed some volumes for the affected users. Fixed, re-running for the affected 10 users.
  • EP-SFT: ROOT releases now properly replicated (and on individual volumes).
There are minutes attached to this event. Show them.
The agenda of this meeting is empty