Ceph » rbd

When restarting interrupted sync (signified by the "end" non-primary snapshot with last_copied_object_number > 0), preserve the "start" non-primary snapshot until the sync is completed, like it would have been done had the sync not been interrupted. This would ensure that the same m_local_snap_id_start is passed to scan_remote_mirror_snapshots() and ultimately ImageCopyRequest state machine on restart as on initial start.

The current behavior goes back to https://github.com/ceph/ceph/commit/281af0de86b17244a2c64af80db0bf84de6d9819 where pruning of unnecessary non-primary snapshot was introduced, but that commit has proven to be buggy.