Hi,
I have this little three node PVE/Ceph cluster.
And due to performance issues I got around to swap my Ceph OSD SSDs once again. I outed one OSD in one of the nodes and it Ceph started rebalancing/remapping/backfilling (as expected). After rebalancing/remaping/backfilling was done, I stopped the OSD and - I think - nothing happened (or maybe it did some more rebalancing/backfilling - not sure). Anyways, I waited a couple of hours and there was no further action, the Ceph circle was all green.
Now when I destroyed the OSD, Ceph started rebalancing/remaping/backfilling again. That was unexpected. I thought that after an OSD was stopped and the cluster has done its thing, it would not rely on it anymore (so that it would not care about the destruction of such OSD anymore).
Can someone help me out why it is behaving like it is?
Thanks!
I have this little three node PVE/Ceph cluster.
And due to performance issues I got around to swap my Ceph OSD SSDs once again. I outed one OSD in one of the nodes and it Ceph started rebalancing/remapping/backfilling (as expected). After rebalancing/remaping/backfilling was done, I stopped the OSD and - I think - nothing happened (or maybe it did some more rebalancing/backfilling - not sure). Anyways, I waited a couple of hours and there was no further action, the Ceph circle was all green.
Now when I destroyed the OSD, Ceph started rebalancing/remaping/backfilling again. That was unexpected. I thought that after an OSD was stopped and the cluster has done its thing, it would not rely on it anymore (so that it would not care about the destruction of such OSD anymore).
Can someone help me out why it is behaving like it is?
Thanks!