Comment 7 for bug 1796886

I think the interesting part of the bug is: "the clients get an authentication failure and never retry thus they sit there unusable until they are manually restarted."

This seems to imply that the config for the amqp client (oslo) hasn't been updated or the service wasn't restarted to pick it up. i.e. if they sit in a failed state, then there's still something wrong on the client code (potentially) or in the data provided to the client. If the cluster comes up and all of the configs on the clients are correct and it *still* isn't working, then it implies that the client wasn't restarted.