We are experiencing some issues with the bucket resharding queue in Ceph Mimic at one of our customers. I suspect that some of the issues are related to the upgrades of earlier versions of the cluster/radosgw. 1) When we cancel the resharding of a bucket, the bucket resharding entry is removed from the queue and almost immediately re-added. We confirm the removal by listing the omapkeys of all the reshard.0000## objects. The relevant omap key is temporarily removed. After a short time it is re-added. I haven't yet determined which process adds it back, but I can only think it is one of the two rgws. 2) We see a lot of objects,1265, in the reshard_pool: "default.rgw.log:reshard". Most look like they might be old to buckets markers or something like that. Most of these objects have not been touched in a long time (mtime 2018). Some numbers: Total buckets: 569 Objects in index pool: 649 Objects in default.rgw.log:reshard namespace: 1265 (of which 16 are the ‘rgw_reshard_num_logs’ objects) these are size '0' Buckets in reshard queue: 41 The objects in "default.rgw.log:reshard" have a similar naming schema to the index objects or , but I cannot relate them directly. Obviously there are loads more than the bucket index objects. The pools used for RGW are for an older generation of Radosgw: .rgw.root default.rgw.control default.rgw.data.root default.rgw.gc default.rgw.log default.rgw.users.uid default.rgw.users.keys default.rgw.buckets.index default.rgw.buckets.data default.rgw.users.email default.rgw.buckets.non-ec default.rgw.users.swift default.rgw.usage My two main questions are: 1) What process, other than dynamic resharding, could cause the re-adding of these buckets to the resharding queue? 2) Do more people see lots of objects in the reshard pool/namespace and can somebody help me understand what these object are? If somebody can point me in the direction of some more documentation or a good talk regarding the resharding mechanism that would also be great. Thanks and with kind regards, Wout 42on