Prometheus

Observability features are available only in Gloo Edge Enterprise

This functionality is turned on by default, and can be turned off a couple of different ways: through Helm chart install options; and through environment variables.

Default Installation

Gloo is shipped by default with prometheus 11.x charts, and provides a set of default values that are not suitable for production usage in most cases. Please provide your own instance of prometheus or configure the helm chart options with production values

For a complete set of options, refer to the Prometheus community Helm charts, or run helm show values prometheus-community/prometheus.

Run Your Own Prometheus

A common setup may be to run your own prometheus, separate from the gloo-provided one. The Gloo Edge Enterprise UI makes use of its own grafana to display dashboards for Envoy and Kubernetes, leveraging gloo custom resources such as Upstreams. You can point gloo’s system grafana toward your prometheus by overriding grafana’s datasources tag, i.e.

grafana:
  datasources:
    datasources.yaml:
      apiVersion: 1
      datasources:
        - name: gloo
          type: prometheus
          access: proxy
          url: http://{{ your.prometheus }}:{{ your.port }}  # fill this in!
          isDefault: true

Find Envoy’s Prometheus Stats

The envoy pod publishes its fairly comprehensive metrics on port 19000. You can view the available ones by running:

# Port-forward to envoy's admin port:
kubectl -n gloo-system port-forward deployment/gateway-proxy 19000

curl http://localhost:19000/stats/prometheus

You can use these to customize the dashboards that get created for every upstream, as described here.