Can this be assigned and triaged?
First, although likely rare, it is not a bug for a bucket that does not exist to be on a reshard list. This can happen when the bucket is added to the reshard list and then removed before the reshard occurs. However two PRs addressed this issue in two ways, both of which are missing from ceph-4.0-rhel-patches (ceph-14.2.4-104). The two PRs are: https://github.com/ceph/ceph/pull/31271 backport to mimic still pending backport to nautilus merged 4 days ago https://github.com/ceph/ceph/pull/31323 backport to mimic still pending backport to nautilus merged 4 days ago So we should cherry-pick them into ceph-4.0-rhel-patches. Eric
I should also add the following. The first of the two PRs -- https://github.com/ceph/ceph/pull/31271 -- allows the user to remove those entries. That did make it into 3.3 (ceph-12.2.12-81). So that does represent a regression. The second PR -- https://github.com/ceph/ceph/pull/31323 -- auto-cleans those entries during reshard queue processing. That is not in 3.3, but would be nice to get into 4.0 as it removes the need for user/amdin intervention. Eric
Matt, Requesting an OK to cherry-pick one or both of the fixes into ceph-4.0-rhel-patches. Details in the previous two comments (#8 and #9). Eric
Thanks, Matt. I've cherry-picked them both.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:0312