Basic Overview

Submariner provides a number of Prometheus metrics, and sets up ServiceMonitor instances which allow these metrics to be scraped by an in-cluster Prometheus instance.

The following metrics are exposed currently:

  • submariner_gateways: the number of gateways in the cluster;

  • submariner_connections: the number of connections to other clusters, with the following labels:

    • local_cluster: the local cluster name
    • local_hostname: the local hostname
    • remote_cluster: the remote cluster name
    • remote_hostname: the remote hostname
    • status: the connection status (“connecting”, “connected”, or “error”)

OpenShift setup

OpenShift 4.5 or later can automatically discover the Submariner metrics. This currently requires enabling user workload monitoring; see the OpenShift documentation for details.

Prometheus operator

To start monitoring Submariner using the Prometheus operator, Prometheus needs to be configured to scrape the Submariner operator’s namespace (submariner-operator by default). The specifics depend on your Prometheus deployment, but typically, this will require you to:

  • add the Submariner operator’s namespace to Prometheus’ ClusterRoleBinding;

  • ensure that Prometheus’ configuration doesn’t prevent it from scraping this namespace.

A minimal Prometheus object providing access to the Submariner metrics is as follows:

kind: Prometheus
  name: prometheus
    prometheus: prometheus
  replicas: 1
  serviceAccountName: prometheus
  serviceMonitorNamespaceSelector: {}
      name: submariner-operator