AFS weekly meeting

Monday 7 Nov 2016, 14:00 → 15:00 Europe/Zurich

31/1-012 (CERN)

AFS service/operation meeting

AFS operations meeting 2016-11-07

present: Dan, Kuba, Massimo, Jan

• CHEP/HEPIX period - no major issues
  • two VMs cold-migrated, while still being alarmed. 2x Operator calls. RFE to Cloud people to use ROGER before/after migration
  • future: new Cinder version should allow to life-migrated even VMs with attached volumes (is already the case for Nova volumes, i.e booted from external volume)
  • running out of "work" space (mails about "BIG trouble.." from pool_monitor) - volumes reshuffled (less space for CMSD, more "full" work servers. Have no tool that will cleanup leftover volumes of the "wrong" type after a partition has been re-tagged.
  • HEPIX summary
• last week: AFS257 instability. was running unmodified upstream 1.6.15 (stable since months, had production data (mix of work,. SVN/GIT, CMS, LHCb..)). Kernel update brought in (via kernel module) 1.6.18-cern, rather unstable. Downgraded, emptied the machine
• migration: external sites ticketed via GGUS if blocking callback, second message was to make them aware of the phaseout
  • will have "no external AFS access"-days to flush more use cases, announced at ITSSB, WLCG etc.
• meeting with ATLAS T0 - Luc - ATLAS-T0 job logs might move elsewhere in January
  • also: underlying volume seems again to back up OK.
• Meeting with Stefan Lueders (triggered by ITSM minutes, access to personal data). 3rd-party access policy is to get explicit OK from Stefan if not in ~/private/ (which needs DG OK). Remove "clone" access for sec team. Keep quarantined volumes and last backups until further notice (Spring 2018 EU data privacy laws?). Reminder about ata access to any file - only if linked to tickets/alarms.
• AuriStor started mailing people who had registered for downloading their software - MacOS Sierra client is available.
• ABS / CASTOR - needs updating to use correct serviceclass "backup" (instead of "afsbackup", is an alias)