Comment 4 for bug 1802640

The timeout always happens as we are waiting for the response from ovsdb-server incrementing next_cfg. The time between successfully calling OVSBridge.create() and set_secure_mode() (which fails) in one of the logs I looked at is 0.3 seconds and there is nothing weird in ovsdb-server.log (we just see the disconnect from the agent after the TimeoutException), so it seems unlikely that things slowed down by over 30x in that short time to cause an ovsdb timeout.

We're not adding any interfaces in the set_secure_mode call and the exception is a TimeoutException so it's definitely hitting here: https://github.com/openstack/ovsdbapp/blob/master/ovsdbapp/schema/open_vswitch/impl_idl.py#L59. So it looks like vswitchd_has_completed(next_cfg) never becomes True. I don't see anything in the ovs-vswitchd logs that would indicate it is having an issue. pre_commit() is definitely being called to increment next_cfg(), so to me it looks like maybe python-ovs is hitting an exception and eating it? It would help to turn on the python-ovs vlog debug logging to see what is going on under the hood. (we normally have it turned off because it is *extremely* chatty.