Recovering from '_nova_to_osvif_vif_binding_failed'

I was showing off my OpenStack dev cluster to a co-worker, and spouting off about how easy and user-proof live-migration is. I challenged him to try it, and he subsequently broke my OpenStack, by trying to live-migrate instances to a compute node targets whose FQDN doesn’t exist (using client CLI commands). Apparently this makes nova sit in a permanent “migration” state (see bug #1643623).

This is a companion discussion topic for the original entry at

Deleting instance wasn’t an option for me, but the following did the trick:

openstack port set --binding-profile vif_type=bridge <port uuid>
openstack server stop <VM uuid>
openstack server rebuild <VM uuid>

Find the port uuid by doing an openstack port list on the instance.

That’s a cleaner fix than mine :slight_smile: