How to monitor AWS RabbitMQ (Amazon MQ service)

[AndresPineros]

I have a cluster-mode RabbitMQ deployment but I'm still getting a single endpoint for https and amqps in AWS. What I understand is that most exporters actually need to get to the instance level and get metrics per each process of a cluster. In this case, if I have a RabbitMQ cluster with 3 nodes, I'd need 3 exporters. Another alternative is that the exporter offers the multi-target scraping pattern.

Anyways, with this information in mind, would this exporter work for an AWS cluster if all the instances are "hiding" behind a single endpoint? Both the amqps:// and https:// endpoints seem to point to a loadbalancer that distributes requests across the 3 instances, based on a nslookup call.

What's the best approach here?

Thanks.

[kbudde]

It is correct this exporter expects to be connected only to one node; otherwise you will have gaps (#131)

Current workarounds:

• Drop the label self from gathered metrics (Exporter is losing data from time to time #131 (comment))
• revert to v0.29
• wait for Feature: RABBITMQ_CONNECTION mode: loadbalancer #230 *

*At the moment, there is a lack of time and priority for this change. Current versions of RabbitMQ offer a Prometheus plugin itself.