Im new to ceph and I have a question related to active+remapped+backfilling and misplaced
Recently I copied more than 10 million objects to a new cluster with 3 nodes and 6 osds
during this migration one of my OSDs got full and health check became ERR I dont know why
but ceph started to write every object on only one osd(can I change this behaviour) and
after it gots full I try to reweight it by utilization and increase the pgs for 1 pool.
cluster became accessible again with status warning and recovery started. I checked the
cluster status for 2 days and I found that I always have 1 pg with status
active+remapped+backfilling and more than 5% misplaced objects. I thought recovery process
takes more days so I leave the cluster to do the recovery in background, till now I have
more active+remapped+backfill_wait pgs and more misplaced objects ( about 10%)
the question is what should I do ? waiting for recovery to finish ? can I speedup this
these servers are in production environment am I in trouble or not ?
Show replies by date