Comment 13 for bug 1463802

Bogdan, not sure where you took that queue name from, but I meant "reply_f7cac1a2428d414bb8b9e0a61291a468".

As you can see:

1) http://xsnippet.org/360736/raw/
2) http://xsnippet.org/360737/raw/
3) http://paste.openstack.org/show/281014/

it *does not* exist at the time nova-conductor tries to consume messages from it.

What we know, is that it didn't survive a restart of the last RabbitMQ server (as it's not *durable*) What we don't know, is why oslo.messaging reconnect detection layer hasn't recreated the queue after RabbitMQ failover (what we see in the synthetic tests).