I'm not sure if that applies since no OSDs were deployed, they were only activated. The node does match a spec file for OSDs, yes. But I didn't expect it to activate the down OSDs. Or has ceph-volume "evolved" to be able to do that? When I reinstalled a host with Quincy (17.2.7), the OSDs were not automatically activated either. So I assume it's something new with Reef. I just scanned the release notes again, but couldn't find anything related to OSD activation. I mean, I don't complain, I think it's great! :-) But for the docs it could be valuable to know when to execute the activation. I haven't done that in Squid yet. Zitat von Cedric <yipikai7@gmail.com>:
Could it be related to automatic OSD deployment ?
https://docs.ceph.com/en/reef/cephadm/services/#disabling-automatic-deployme...
On Fri, Feb 14, 2025, 08:40 Eugen Block <eblock@nde.ag> wrote:
Good morning,
this week I observed something new, I think. At least I can't recall having seen that yet. Last week I upgraded a customer cluster to 18.2.4 (everything worked fine except RGWs keep crashing [0]), this week I reinstalled the OS on one of the hosts. And after a successful registry login, the OSDs were automatically activated without me having to run 'ceph cephadm osd activate <host>' as documented in [1]. Zac improved the docs just last week, is that obsolete now? Or is that version specific? A few weeks ago we reinstalled our own Ceph servers as well, but we still run Pacific 16.2.15, and there I had to issue the OSD activation manually. Can anyone confirm this?
Thanks! Eugen
[0] https://tracker.ceph.com/issues/69885 [1]
https://docs.ceph.com/en/latest/cephadm/services/osd/#activate-existing-osds _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io