Hi,
Does somebody see PG inactive like this before?
We get first pool outage:
PG_AVAILABILITY Reduced data availability: 2 pgs inactive
pg 4.1f1 is stuck inactive for 8637.783533, current state clean+premerge+peered, last acting [312,358,331]
pg 4.9f1 is stuck inactive for 8637.783331, current state remapped+premerge+backfilling+peered, last acting [312,331,374]
Then added alerts for premerge state for long time, and get second outage:
PG_AVAILABILITY Reduced data availability: 2 pgs inactive
pg 4.1d9 is stuck inactive for 1000.400328, current state remapped+premerge+backfilling+peered, last acting [328,315,352]
pg 4.9d9 is stuck inactive for 1000.400333, current state remapped+premerge+backfill_wait+peered, last acting [328,315,352]
Actually, from Nautilus we was made PG reduce many times. This is first time problem occurence and only on one cluster
Before this, PG reducing was going for a week on this cluster, but only one by one PG and ignore max_misplaced option totally - I wasn't debug why
For a 3 hour before first outage - osd.362 was added to this pool
Tracker for this - https://tracker.ceph.com/issues/52509 <https://tracker.ceph.com/issues/52509>
Thanks,
k
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
