AFS operations meeting 2015-01-05
present: Dan, Massimo, Kuba, Jan
Issues
- quiet Xmas period:
- no issues seen with backup/ABS (Kuba checked logs to make sure)
- afs289 got BOINCed (or at least segfaulted) after "work.boinc access rescheduled"
- afsacl machines ran out of memory as usual (and afsalc01 also out of disk, after too many kernel crash dumps = old 1.6.6-2cern. updated.)
- Service: ABS check now keeps most logs local (and logrotates), in order not to fill up /p/var. Will eventually open SNOW tickets for missing backups.
- Upcoming:
- investigate using "xrdcp"/"xrdcopy" for the ABS backups, is already installed. Should make us (more) independent of CASTOR nameserver overloads (xrootd uses the stager-local nsd), but need to see whether copying from STDIN is supported.
- "disaster recovery" document, take 2 (is using AFS as example)
- "EOS / CERNBOX" as future home/work/project space alternative, need a (meta)plan..
There are minutes attached to this event.
Show them.