Comment 9 for bug 1490051

Some brain dump, I think I am onto smth.

So we saw another failure today: http://logs.openstack.org/02/216902/4/check/gate-neutron-dsvm-functional/a97df90/console.html

Since we reuse rootwrap daemons per worker, and the error looks like the daemon does not see the functional tests filters, the first thing I checked is which test started the daemon for the worker. It's:

http://logs.openstack.org/02/216902/4/check/gate-neutron-dsvm-functional/a97df90/logs/neutron.tests.functional.agent.linux.test_interface.OVSInterfaceDriverTestCase.test_plug_checks_if_bridge_exists_vsctl_.log.txt.gz

Note that it started at 05:44:47.284.

Now compare it to console log where we call to deploy_rootwrap.sh at 05:44:47.335. So it looks like the test started to execute before tox called to deploy_rootwrap, and the daemon was started before the filters were deployed.

Now, I wonder how tests could start at the time when we still have 'dsvm-functional create: /opt/stack/new/neutron/.tox/dsvm-functional' in console log. The test relies on testscenarios. Could it be an issue?