Comment 14 for bug 1483182

So the problem was seen since 6.1 release, there were no changes in packages between 6.0 and 6.1, what was changed is instead of bridge, to connect services (in containers) we use a single interface.

Most likely the problem is related to temporary connectivity problem, how it can happen, when all services are connected to localhost is not found.

Similar symptoms can be seen by dropping packages to RabbitMQ from Astute, there is no attempt to reconnect and Astute successfully sends the data to RabbitMQ, kernel doesn't return error on writev into socket. Actually there is an attempt to reconnect in an hour, it can be seen in attached logs, to be precisely in 50 minutes, which is standard heartbeat timeout in RabbitMQ. It was reduced, so now if there is problem in connectivity we will see the problem much earlier and try to perform reconnect.