Hi!

I know there is a way to alter the database and hard reboot the instances to migrated it to another hypervisor after another crashed, but is there also a way to pause / resume the vms instead of hard rebooting?

In my case let's assume I have a shared storage containing all vms and only the openstack-nova-compute service died but libvirtd and the machines are still running. I tried to pause them with "nova suspend <uuid>", but now they are in suspending task and cannot be resumed. Next I tried to suspend them with "virsh suspend <uuid>" on the old hypervisor, altered the db and tried to resume them with nova on the new compute node, but unfortunately nova didnt recognize the vms got suspended at all. If I would have shut them down using virsh nova gets a status update.

Am I missing something or is this way of migrating just not possible?

Thanks & Greets

Basti