In Kubernetes application metrics are typically scraped from each Pod and enriched with Kubernetes metadata (Namespace name, Pod name, container name etc.) by Prometheus. Thus application components need to expose application specific metrics in Prometheus format on a metrics endpoint (by default named /metrics). Application components can then inform Prometheus about their metrics endpoint using the annotation prometheus.io/path.
A metrics endpoint is only required, if an application component wishes to monitor certain application-specific metrics. Generic metrics about resource usage and potential errors are collected automatically at cluster level.
An NGINX deployment with a sidecar container which exports metrics on port 9113 on the default /metrics endpoint. The deployment tells Prometheus about this endpoint using the annotations prometheus.io/path and prometheus.io/port.
Use the following command to create the Deployment:
kubectl apply -f examples/Check the pod using
kubectl get pods -o wideConfirm that the Pod exposes a metrics on port 9113 endpoint by running the following command:
kubectl run curl --image=curlimages/curl --rm -it -- curl <Pod IP>:9113/metricsUse the following command to delete the Deployment:
kubectl delete -f examples/