AFS weekly meeting

Monday 6 Jul 2015, 14:00 → 15:00 Europe/Zurich

31/1-012 (CERN)

AFS service/operation meeting

AFS operations meeting 2015-07-06

present: Kuba, Dan, Jan

Issues:

• /p/var full (ABS logs) temporarily extended (apparently somebody else had already done this in the past? was already 200GB), eventually need proper logrotate (or sub-voluming). Affected /p/etc/rcsedit..
• some automatic volmove failures, being looked at.
• while restoring volumes for sec team noted that sometime get python crashes due to half-restored disk files. usually OK when retried.
• when asked for new AFS projects, systematically propose EOS space in addition (assign ticket to EOS team, propose 10x AFS space). Can still give AFS space, but marked as for the short term.
• fix for "AFS stuck at shutdown" has most likely been found by Linux.Support (for 1.6, CERN had renamed the init script ("service"), but not renamed lockfile inside -> init script not executed at shutdown). Need new release (ideally on 1.6.12).
  • drop CERN-modification altogether for 1.6? No, at least the "prevent getting stuck without network"-patch is required for SLC6
• Wigner:
  • afsdb4/afsdb5 are now advertised via DNS. CERN puppet machines should see these (as they run with "-adfsdb", most other clients use CellServDB, and the new machines are not in there). And should have proper "fs setserverprefs", so these machines will not accidentally cross-talk
  • afs500/afs501 will get some readonly replicas (manually placed, but __choose_disk has been patched to honour a new "readonly" tag).