I have a docker instance of haproxy in front of a 3 node rabbitmq cluster.
I the same Docker swarm I have a Springboot microservice that access the queue thru the proxy.
If I let everything come up on its own the microservice keeps trying to connect to rabbitmq and cannot.
If I restart the haproxy docker container, when it comes up everything is fine.
This makes it look like either 1) if Haproxy cannot connect to the rabbitmq servers because they are not up, it does NOT eventually connect to them when they are up
or 2) after trying to connect thru haproxy and failing, a restart of haproxy makes them try again and succeed.
neither makes sense to me. surely if haproxy is looking for 3 servers but one goes down, when it comes back up it will eventually pull it into the round robin?
Can anyone explain what (might be) happening?
