Send broker metrics to Grafana Cloud (Mimir)

Prometheus all the way down

Grafana Cloud keeps your metrics in Mimir, a Prometheus-compatible time-series database. Mimir expects data to arrive via remote write, and you query it with PromQL. Both ends of this integration speak Prometheus natively.

That matters because CloudAMQP already exposes broker and server metrics in Prometheus format. None of it gets remapped into a vendor-specific model along the way. The metric names and labels in our documentation are exactly what you type into Grafana. That also means your existing dashboards and alert rules will continue to work, since they were already written against Prometheus metrics.

What you can export

You choose what leaves your cluster. The default set covers both sides of the picture:

• Server metrics: CPU, memory, disk and filesystem usage, network, and paging
• Broker metrics: connections, channels, consumers, queue depth, unacked messages, and publish, deliver, and ack counters

The metrics filter takes a list of metric names, so you send only the series your dashboards and alerts actually use, instead of paying to store all of them. Our documentation lists the default set, and the filter is not limited to it: you can name anything your broker exposes on /metrics and /metrics/detailed , though detailed metrics get expensive on clusters with many queues and connections. Set the filter per integration, in the console or through the API.

Every series carries an instance label with the node name and a cluster label with the cluster name, so a three-node cluster stays readable in one query. Add your own labels as custom tags when you configure the integration, for example env=production . Metrics arrive every 60 seconds, or every 120 seconds on Sassy Squirrel instances.

Open Explore in Grafana, pick your Prometheus data source, and query away:

lavinmq_queue_messages_ready{cluster="your-cluster-name"}
system_cpu_utilization_ratio{instance="your-cluster-name-01"}

Setting it up

1. Sign in to the Grafana Cloud Portal and open your stack.
2. Click Send Metrics on the Prometheus tile. That page holds all three values you need: the remote write endpoint, your numeric instance ID, and a button to generate an API token.
3. Generate a token with the metrics:write scope, or use a service account token with the MetricsPublisher role.
4. Paste the endpoint, instance ID, and token into the Integrations tab for your instance in the CloudAMQP console and save. You can also enable the integration with the Instance API.

The instance ID becomes the basic auth username, and the token becomes the password. One thing worth knowing if you already ship logs to Grafana Cloud: the Prometheus instance ID is not the same number as the Loki one. Copy it from the Prometheus tile, not from your existing Grafana Cloud (Loki) log integration.

Summary

Grafana Cloud is now a first-class metrics destination for CloudAMQP. You get server and broker metrics in Prometheus format, labeled per node and per cluster, pushed every minute via remote write, with a metric filter that trims the export down to the series you actually use. Setup takes an endpoint, an instance ID, and a token.

The Grafana Cloud metrics integration documentation lists the default metrics and their labels. Questions or a destination you would like us to support next? Write to support@cloudamqp.com or browse all CloudAMQP integrations.