On 10/6/2020 4:43 PM, Stefan Kooman wrote:
On 2020-10-06 15:27, Igor Fedotov wrote:
I'm working on improving PG removal in master, see: https://github.com/ceph/ceph/pull/37496
Hopefully this will help in case of "cleanup after rebalancing" issue which you presumably had. That would be great. Does the offline compaction with the ceph-kvstore-tool follows a completely different removal procedure and is that why it's that much faster?
First of all - compaction is primarily about rearranging DB data in an optimal way not removing it. Compaction after massive removals just optimizes DB data layout and eliminates all the leftovers from the previous rewrites/removals that negatively impact DB performance. Offline compaction work in an exclusive access mode and doesn't care about parallel DB operations. Hence it's presumably much more effective comparing to the online one. Not to mention that the latter is rather partial...
@Kristof: https://github.com/ceph/ceph/commit/93e4c56ecc13560e0dad69aaa67afc3ca053fb4c is a commit by Wido that would help to enable compaction at OSD (re)start.
Gr. Stefan