README.En US

Introduction

The official AutoMQ documentation describes how metrics are exposed and how to integrate with monitoring systems. For details, refer to AutoMQ.

Recommended approach

Send AutoMQ metrics to an OTel Collector Prometheus exporter, or scrape an AutoMQ Prometheus endpoint directly. The real-data test used AutoMQ 1.5.5 and http://automq:8890/metrics; use the port configured by your deployment.

Create conf/input.prometheus/automq.toml:

interval = 15

[[instances]]
urls = [
  "http://<automq-or-otel-collector>:8890/metrics"
]

url_label_key = "otel_collector"
url_label_value = "{{.Host}}"
labels = { source = "automq" }

Do not set url_label_key to instance: AutoMQ already emits an instance label. Do not overwrite the original job label either. The dashboards use these labels for the cluster, node, and active-controller variables.

curl -fsS http://<automq-or-otel-collector>:8890/metrics \
  | grep -E 'process_runtime_jvm_cpu_utilization_ratio|kafka_request_count_total' \
  | head
./categraf --test --inputs prometheus

Topic, consumer-group, and object-storage panels require real message and object-store traffic.