test_update_access_server_address failing in check-tempest-dsvm-postgres-full

Didn't see an existing bug for this but needed to recheck against it:

http://logs.openstack.org/03/57003/11/check/check-tempest-dsvm-postgres-full/5a6a05c/console.html

2013-12-05 01:01:31.307 | Traceback (most recent call last):
2013-12-05 01:01:31.307 | File "tempest/api/compute/servers/test_servers.py", line 94, in test_update_access_server_address
2013-12-05 01:01:31.307 | resp, server = self.create_test_server(wait_until='ACTIVE')
2013-12-05 01:01:31.308 | File "tempest/api/compute/base.py", line 117, in create_test_server
2013-12-05 01:01:31.308 | server['id'], kwargs['wait_until'])
2013-12-05 01:01:31.308 | File "tempest/services/compute/json/servers_client.py", line 160, in wait_for_server_status
2013-12-05 01:01:31.308 | extra_timeout=extra_timeout)
2013-12-05 01:01:31.308 | File "tempest/common/waiters.py", line 73, in wait_for_server_status
2013-12-05 01:01:31.309 | raise exceptions.BuildErrorException(server_id=server_id)
2013-12-05 01:01:31.309 | BuildErrorException: Server aea6135c-a998-470e-b2e1-d8794f7805bf failed to build and is in ERROR status

I do see this in the nova scheduler log:

2013-12-05 00:41:45.588 INFO nova.filters [req-29e72655-eb62-44a9-8149-e031de33a85f ServersTestJSON-tempest-2138013281-user ServersTestJSON-tempest-2138013281-tenant] Filter ComputeFilter returned 0 hosts
2013-12-05 00:41:45.588 WARNING nova.scheduler.driver [req-29e72655-eb62-44a9-8149-e031de33a85f ServersTestJSON-tempest-2138013281-user ServersTestJSON-tempest-2138013281-tenant] [instance: aea6135c-a998-470e-b2e1-d8794f7805bf] Setting instance to ERROR state.

Looks like the ComputeFilter couldn't find a host so it bombs out. I'm not sure why the timestamps don't line up with the failure in console.html.

Looks like this has been showing up since at least 11/22:

http://logstash.openstack.org/#eyJzZWFyY2giOiJtZXNzYWdlOlwiRkFJTDogdGVtcGVzdC5hcGkuY29tcHV0ZS5zZXJ2ZXJzLnRlc3Rfc2VydmVycy5TZXJ2ZXJzVGVzdEpTT04udGVzdF91cGRhdGVfYWNjZXNzX3NlcnZlcl9hZGRyZXNzW2dhdGVdXCIgQU5EIGZpbGVuYW1lOlwiY29uc29sZS5odG1sXCIiLCJmaWVsZHMiOltdLCJvZmZzZXQiOjAsInRpbWVmcmFtZSI6ImFsbCIsImdyYXBobW9kZSI6ImNvdW50IiwidGltZSI6eyJmcm9tIjoiMjAxMy0xMS0yMFQwMjoyMDoxMyswMDowMCIsInRvIjoiMjAxMy0xMi0wNVQwMjozNToxMyswMDowMCIsInVzZXJfaW50ZXJ2YWwiOiIwIn0sInN0YW1wIjoxMzg2MjExNjY5NDEyfQ==

The ComputeFilter itself hasn't changed since 8/21, so nothing interesting there.

This nova scheduling fix merged on 11/21 but I'm not sure why/how that would cause this:

https://github.com/openstack/nova/commit/1d769b0c37f24028bceae568ca66bb0426b28b69