First, I would restart the active mgr, the current status might be outdated, I've seen that lots of times. If the pg is still in remapped state, you'll need to provide a lot more information about your cluster, the current osd tree, ceph status, the applied crush rule etc. One possible root cause is that crush gives up too soon, this can be adjusted (increase set_choose_tries) in the rule. Zitat von Jorge Garcia <jgarcia@soe.ucsc.edu>:
We were having an OSD reporting lots of errors, so I tried to remove it by doing:
ceph orch osd rm 139 --zap
It started moving all the data. Eventually, we got to the point that there's only 1 pg backfilling, but that seems to be stuck now. I think it may be because, in the process, another OSD (103) started reporting errors, too. The pool is erasure k:5 m:2, so it should still be OK. I don't see any progress happening on the backfill, and ceph -s has been reporting "69797/925734949 objects misplaced (0.008%)" for days now. How can I get it to finish the backfilling, or at least find out why it's not working?
ceph version: quincy
Thanks!
Jorge _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io