Nova delete instance fails if cleaning is not async

[Yuriy Zveryanskyy]

When node cleaning is not async (conductor starts cleaning immediately in do_node_tear_down()) Nova delete instance fails. Nova tries to update Ironic port (remove vif id), but conductor releases a lock only after thread with cleaning completed. Because cleaning usually is not a short time action Nova fails after retries. This bug can affects any out-of-tree driver, reproduced on Ansible deploy driver PoC ( https://review.openstack.org/#/c/238183/ ).

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/275080

[Sean Dague (sdague)]

It is not clear that this is a nova issue, but a wholely ironic issue.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Yuriy Zveryanskyy (<email address hidden>) on branch: master
Review: https://review.openstack.org/275080

[Dmitry Tantsur (divius)]

Can someone please retest it? I've seen some fixes in Nova (and probably Ironic) around this code.

[Launchpad Janitor (janitor)]

[Expired for Ironic because there has been no activity for 60 days.]