@Armando Looking through the stack traces I linked in comment #1 ([2] and [3]) seem to support your suspicion. It looks like prepare_network is shared among a lot of different tests and is run near the beginning of the test.
I wonder if the two failure modes, [2] and [3], can be examined independently. I think we should start focusing on failures containing a stack trace that looks like [2]. These seem to always be accompanied by a failure in the delete_wrapper which appears to be a problem with cleaning up resources, maybe from the previous test? The failure mode in [2] also seems to be much more common than the one in [3].
@Armando Looking through the stack traces I linked in comment #1 ([2] and [3]) seem to support your suspicion. It looks like prepare_network is shared among a lot of different tests and is run near the beginning of the test.
I wonder if the two failure modes, [2] and [3], can be examined independently. I think we should start focusing on failures containing a stack trace that looks like [2]. These seem to always be accompanied by a failure in the delete_wrapper which appears to be a problem with cleaning up resources, maybe from the previous test? The failure mode in [2] also seems to be much more common than the one in [3].
[1] http:// paste.openstack .org/show/ 455392/ paste.openstack .org/show/ 455397/ paste.openstack .org/show/ 455399/
[2] http://
[3] http://