Skip to main content

Unmanaged Prometheus/Grafana

Pre-Requisites​

  1. Create a Kubernetes cluster.
  2. Install Retina DaemonSet (see Quick Installation).

Configuring Prometheus​

  1. In this example, we will install Prometheus via the community supported helm chart. If you already have this chart deployed, skip to step 3.
helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm repo update

  1. Save these Prometheus values below to deploy/prometheus/values.yaml

  2. Install the Prometheus chart

helm install prometheus -n kube-system -f deploy/prometheus/values.yaml prometheus-community/kube-prometheus-stack

Or if you already have the chart installed, upgrade how you see fit, providing the new job name as an additional scrape config, ex:

helm upgrade prometheus -n kube-system -f deploy/prometheus/values.yaml prometheus-community/kube-prometheus-stack

Note: Grafana and kube-state metrics may schedule on Windows nodes, the current chart doesn't have node affinity for those components. Some manual intervention may be required.

  1. Verify that the Retina Pods are being scraped by port-forwarding the Prometheus server:
kubectl port-forward --namespace kube-system svc/prometheus-operated 9090
  1. Then go to http://localhost:9090/targets to see the Retina Pods being discovered and scraped:

alt text

Configuring Grafana​

Create a Grafana instance at grafana.com and follow Configuring Grafana, or use the one installed from above.

  1. Get the Grafana password:
kubectl get secret -n kube-system prometheus-grafana -o jsonpath="{.data.admin-password}" | base64 --decode ; echo

  1. Import the new dashboard from here.

  2. Metrics should be visible:

alt text