RabbitMQ prometheus exporter granularity control

[Dachi Mshvidobadze]

Hello!

When enabling per-object metrics in the rabbitmq_prometheus plugin, it produces very granular metrics (too granular even) - but from what I've found that is not very adjustable.

For instance, metric `rabbitmq_queue_messages_published_total` is very useful but it comes with the `channel` label, which raises cardinality for these metrics, and with a very big and active cluster (1k-10k connections per cluster node), it gets completely out of hand.

I was wondering if anyone else had the same issue, and how did they deal with it?

[Michal Kuratczyk]

You can check /metrics/detailed to see if that meets your needs: https://github.com/rabbitmq/rabbitmq-server/blob/main/deps/rabbitmq_prometheus/metrics-detailed.md

Either way, I agree that some metrics could probably be dropped or exposed differently. As always the problem is that given the large and diverse RabbitMQ user base, whenever we think something could be changed/removed, it usually turns out there is a group of users for whom it's critical... :)

In this specific case, you can always drop or relabel metrics on the Prometheus level.

Best,

--
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/b44f6d05-c765-40a7-b990-f30aa0c743a5n%40googlegroups.com.

--

Michał

RabbitMQ team

RabbitMQ team

[Dachi Mshvidobadze]

Prometheus level it is I guess :)

Thanks.