.net core 2.0
Our saga was configured to use sql server for the transport layer, but when we changed to rabbitmq we have found that a single uncaught except can block our saga queue. This has made our application unscalable.
We are using conventional routing topology, have set message processing concurrency on the endpoint to 1, a request timeout of 5 seconds (when sending messages) and have gone with the default for transport transaction mode and number of retries etc. Is there anything obvious in our configuration which would be causing this? I would have thought that nservicebus would be able to cope with uncaught exceptions or do we have to write a wrapper to make sure that no exception is uncaught?