Possible to get stuck libvirt entry that can't be destroyed

[nh2]

I once cancelled a nixops deploy to a libvirt machine (with nixops being run on NixOS) with a plain kill when it was stuck (I had launched it via ssh without -t so I wasn't seeing any output).

Afterwards, I could no longer destroy, it was stuck:

[test@machine:~/configs/nixops]$ time nixops destroy -d test-libvirt
node-1> destroying... 
node-3> destroying... 
node-2> destroying... 
node-1> shutting down... 
node-3> shutting down... 
node-2> shutting down... 
node-1> Domain nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-1 destroyed
node-1> 
node-3> Domain nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3 destroyed
node-3> 
node-2> Domain nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-2 destroyed
node-2> 
node-3> error: failed to get domain 'nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3'
node-3> error: Domain not found: no domain with matching name 'nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3'
node-3> 
node-1> Domain nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-1 has been undefined
node-1> 
node-2> Domain nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-2 has been undefined
node-2> 
error: command ‘['virsh', '-c', 'qemu:///system', 'undefine', u'nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3']’ failed on machine ‘node-3’ (exit code 1)

Trying to redeploy:

[test@machine:~/configs/nixops]$ time nixops deploy -d test-libvirt 
node-3> starting...
node-3> error: failed to get domain 'nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3'
node-3> error: Domain not found: no domain with matching name 'nixops-9317c270-d450-11e7-9b83-001e67b6635e-node-3'

Trying to delete:

$ time nixops delete -d test-libvirt
error: cannot delete this deployment because it still has resources

[nh2]

I managed to get rid of it with delete --force, and recreated the deployment as a workaround.

It would still be useful though to know if the above should happen on SIGTERM or not.