x-overflow = reject-publish-dlx on quorum queues

[Sabina Rosca]

Hi everyone,

I have a question about using queue argument x-overflow with value reject-publish-dlx on quorum queues.

The quorum queues feature matrix and migration guide mention that overflow dead lettering is not supported for quorum queues.

But in practice this doesn't actually seem to be the case.

When testing with RabbitMQ versions 3.10.13 and 3.11.3 (I used Docker containers), I was able to declare a quorum queue with the following arguments:

• x-overflow = "reject-publish-dlx"
• x-max-length = 1
• x-dead-letter-exchange = "" (empty string for the default exchange)
• x-dead-letter-routing-key = "MyDLQ"

The queue declaration didn't throw any exceptions, and when publishing messages to this quorum queue, all messages after the first one were routed to queue MyDLQ, as indicated by the x-dead-letter arguments.

The only difference I could find compared to a classic queue with the same arguments was that when enabling publisher confirms on the channel used to publish messages, I received BasicNack events for the messages published to the classic queue, but did not receive BasicNack for the quorum queue. I'm using the .NET RabbitMQ.Client library, version 6.4.0 btw.

Am I missing something here? Or is the only difference the BasicNack?

Thank you!

[Karl Nilsson]

It is possible that we've missed to validate against that specific value, I will check this.

What most likely happened is that your queue is running in drop-head mode and is why your messages turn up in the dead letter queue.

Cheers

karl

--
You received this message because you are subscribed to the Google Groups "rabbitmq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rabbitmq-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/rabbitmq-users/48fb6f54-7e18-46bb-b30f-db3c79863018n%40googlegroups.com.

--

Karl Nilsson

[Sabina Rosca]

Oh, it defaulted to drop-head, and because I configured the x-dead-letter args, dropped messages were routed to the dead letter queue. That makes sense.

Thank you for the quick response!