kernel corrupted during upgrade, now can't add OSD

Jul 29, 2025
7
0
1
I have a 3 server PVE setup, each with 6 NVME drives, after an upgrade from 8.2 to 8.4, server 2 would get stock at boot up, had to boot to an old kernel 5.X, then I forced reinstall the kernel 6.8.X, now the OSD show out, I've tried removing the OSDs, using all known methods of zapping them (Ceph auth del X, ceph-volume lvm zap, ceph osd purge, dmsetup remove, etc.

No matter what I do, when I try and add the OSD, they don't show up under Ceph > OSD > Host, but show as Ghost OSDs, I can get them to show again if I export the crush map, edited it and import it, but just put me back to square one, the OSD show as out, I can click In but they won't start and after a few minutes take themselves out again.

Right now, all 3 servers are up to date with 8.4.1 with Kernel 6.8.12 and Ceph version 18.2.4 and also a PBS. Each server has 2 10Gbps, one dedicated for Ceph and one for VMs access.

Thank you in advance for your assistance.
 
Last edited: