Open Stack

Howdy all,

FYI we have gate failures of the recently added 
test_*_with_qos_min_bw_allocation tests [1] in the nova-multi-cell job 
on the master, stable/victoria, and stable/ussuri branches. The failures 
occur during cross cell migrations.

I have opened a bug for the failure on the master branch:

* https://bugs.launchpad.net/nova/+bug/1907522

The issue here is that we fail to create port bindings in neutron during 
a cross cell migration in the superconductor:

nova.exception.PortBindingFailed: Binding failed for port <port uuid>

and that corresponds to a failure in the neutron server log where it 
fails the port binding with:

neutron_lib.exceptions.placement.UnknownResourceProvider: No such 
resource provider known by Neutron

I don't yet know what is going on here ^.

For the bug on stable/victoria and stable/ussuri I have opened this bug:

* https://bugs.launchpad.net/nova/+bug/1907511

and have a WIP stable-only patch proposed that needs tests:

https://review.opendev.org/c/openstack/nova/+/766364

I just wanted to see ASAP if the nova-multi-cell job will pass on it.

The issue here ^ is that during a cross cell migration, we aren't 
targeting the cell database for the target host when we attempt to 
lookup the service record of the target host.

For the stable branch failures I think the failure rate is 100% and it 
looks like it might also be 100% for the master branch failures.

Cheers,
-melanie

[1] https://review.opendev.org/c/openstack/tempest/+/694539