@rogeralsing That is basically it. In your example you work with a batch of messages. So in this case if a message fails for on a specific actor I no longer want any messages to process for that given actor. The reason is I want to preserve the order of the messages from the queue based on the hash. I guess I could use a Circuit Breaker just for that specific route and once it gets tripped it would stash the messages along with a supervisor that would not retry if the circuit is tripped. This will probably never be an issue but theoretically a service like a DB could be there one second then gone the next and by the time it returns the message could conceivably timeout.