linkerd2/doc/prometheus.md

+++ title = "Exporting metrics to Prometheus" weight = 5 docpage = true [menu.docs] parent = "prometheus" +++

If you have an existing Prometheus cluster, it is very easy to export Conduit's rich telemetry data to your cluster. Simply add the following item to your scrape_configs in your Prometheus config file:

    - job_name: 'conduit'
      kubernetes_sd_configs:
      - role: pod
        namespaces:
          # Replace this with the namespace that Conduit is running in
          names: ['conduit']
      relabel_configs:
      - source_labels: [__meta_kubernetes_pod_container_port_name]
        action: keep
        regex: ^admin-http$

That's it! Your Prometheus cluster is now configured to scrape Conduit's metrics. Conduit's metrics will have the label job="conduit" and include:

• requests_total: Total number of requests
• responses_total: Total number of responses
• response_latency_ms: Response latency in milliseconds

All metrics include the following labels:

• source_deployment: The deployment (or replicaset, job, etc.) that sent the request
• target_deployment: The deployment (or replicaset, job, etc.) that received the request