Comment 2 for bug 1880509

Yes, that is correct. The host works normally otherwise.

Note, I did have to do some clean up on the Cinder database due to failed migrations during the Queens to Rocky upgrade, and while I'm not intermittently familiar with the Cinder database, this does not appear to be the root cause of the issue because the Volume attaches to a different host running on a different hypervisor host.

Also, Google searches that returned, for example, "HypervisorUnavailable: Connection to the hypervisor is broken on host" were primarily related to libvirt being down. This is not the case here. I did not see anything that related to and had a solution related to the issue in this report.

I also took the steps of reconfiguring both Nova and Cinder but the issue remains.

Finally, in Queens, volumes attached to all hypervisor hosts with issue.