Exceeding maxDeliveryAttempts

Question

As per my understanding default value for maxDeliveryAttempts is 5?
Even after explicitly setting the value of 5 at times I've observed retry to exceed 5 ( upto > 10 recently). Configurations are SetNumberOfWorkers:5,SetMaxParallelism: 10, SimpleRetryStrategy(maxDeliveryAttempts: 5, secondLevelRetriesEnabled: true),

Within handler, if we are unable to process the message as a best-suggested practice should I explicitly throw an exception, return null or just let the exception propagate ( at the moment I catch the exception log and rethrow same) - can this impact the behaviour?

I've also implemented IHandleMessages> to do some house keeping. Note: using rebus with rabbitmq

Answer 1

First off, "best practice" with Rebus is to let exceptions bubble out of the handler and let Rebus do the logging. It will log each failed processing attempt as a WARNING, and then it will log all of the caught exceptions as an AggregateException as an ERROR when moving the message to the error queue.

Unless you enable 2nd level retries :) in that case, Rebus will proceed on to a second round of deliveries after the initial 5 WARNINGs, only dispatching the message as an IFailed<TMessage>.

If the subsequent five delivery attempts also fail (e.g. because there's no handler that implements IHandleMessages<IFailed<YourMessage>>), then you will get an additional 5 WARNINGs in the log, followed by an ERROR, when the message finally gets moved to the error queue.

I hope that makes it clearer :) let me know if there's something I should explain in more detail.