[Suggestion] Message count metrics broken down by queue type

[the-mikedavis]

RabbitMQ series

4.2.x

Operating system (distribution) used

Linux

How is RabbitMQ deployed?

Other

What would you like to suggest for a future version of RabbitMQ?

The default /metrics endpoint currently exposes queue depth gauges (rabbitmq_queue_messages_ready, "_unacked, "_messages) as a single unlabeled sum across all queues on the node. Operators who want to understand capacity by queue type (for example, to alert when quorum queue depth grows independently of stream backlogs) must scrape the expensive /metrics/per-object or /metrics/detailed endpoints and re-aggregate client-side. I think we should calculate and emit a (new? to avoid a breaking change?) metric labelled by queue-type. This would make it clear when you are following the best practices for the queue-type. For example "keep queues short" does not really apply to streams.