Hi there ! Being suspicious about replication between two clusters, I've done a radosgw-admin data sync init on the secondary zone. Since then, after a lot of activity, I'm stuck with recovering shards, nothing moves. Incremental sync still work. Wondering if I had a bad state also on the primary, I also did a data sync init on the primary... And now, it's also stuck with recovering shard ! In sync error list, I can find some "failed to sync bucket instance: (125) Operation canceled" errors I also tried to rewrite some buckets chown in those errors, but nothing changes. Strange, in those errors, the objects names are not real objects, example : "name": "replic_cfn_rec/cfb0047:aefd400 3-1866-4b16-b1b3-2f308075cd1c.20298566.4:11[0]" I wonder what is this ending ":10[0]". I also tried to remove stale instances, but nothing. I've still not retry a data sync init on secondary, perhaps I should, but the generated activity is impactfull. Can we reduce that resync activity priority ? ah, My primary cluster is on Reef 18.2.4, the secondary still on 18.2.2 (needs OS upgrade, Ubuntu 18.04). -- Gilles -- Gilles _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |