Ceph Upgrade to Octopus - Degraded data redundancy: 15/11603845 objects degraded

[mcdowellster]

Hello,

I followed the upgrade guide as I want to move my cluster to PVE7. Post upgrade I'm stuck with 15 objects on a single PG that REFUSE to repair. It's been in a "stuck" state now for hours. There are no slow IO issues. Ceph osd logs look clean. I followed the posted proxmox upgrade guide to the letter. The only thing I've yet to do is restart the physical node it's on. I suppose I could also restart the OSD in question but I fail to see how that would resolve this.

Any input?

 

[mcdowellster]

Found it.

For reference, my cluster only has two nodes with OSDs and three with monitors. The health pool was the problem. It was set to 3/2 instead of 2/2