Comment 1 for bug 1742936

I looked at this with Ihar for a while today and didn't find any smoking gun in the log files.

There are a few things we'd need to do debug this further from the neutron side, assuming it's a networking issue:

1. Stop the system under test when this fails so we can login and look around.

2. Make sure we have the console log of the instance(s). If this isn't doable via some setting we need to look at something like https://stackoverflow.com/questions/12290336/how-to-execute-code-only-on-test-failures-with-python-unittest2 to dump the console on failure

3. Log haproxy and metadata agent serviced requests. The haproxy conf file shows it should log to /dev/log, which is exposed in the container, but I couldn't find any evidence of logging anywhere. The metadata agent does not over-ride debug level like other agents, should set "debug = True" in metadata_agent.ini

In the meantime it might be worth skipping the test until we can gather additional information if it is critical.