Metrics quickstart

This section describes how to enable metrics reporting on the Policy Server.

note

Before continuing, make sure you completed the previous OpenTelemetry section of this book. It is required for this section to work correctly.

We are going to use Prometheus to scrape metrics exposed by the Policy Server.

Install Prometheus

We will use the Prometheus Operator, that allows us to define Prometheus' Targets intuitively.

There are many ways to install and set up Prometheus. For ease of deployment, we will use the Prometheus community Helm chart.

The prometheus-operator deployed as part of this Helm chart defines the concept of Service Monitors, to define which services should be monitored by Prometheus declaratively.

In our case, we are adding a ServiceMonitor targeting the kubewarden namespace for services that match labels app=kubewarden-policy-server-default and app.kubernetes.io/name: kubewarden-controller. This way, the Prometheus Operator can inspect which Kubernetes Endpoints are tied to services matching these conditions.

Let's create the two ServiceMonitors named kubewarden-controller and kubewarden-policy-server to be used by the default prometheus instance installed by the Helm chart. For that, you can create the following values file:

cat <<EOF > kube-prometheus-stack-values.yaml
prometheus:
  additionalServiceMonitors:
    - name: kubewarden
      selector:
        matchLabels:
          app: kubewarden-policy-server-default
      namespaceSelector:
        matchNames:
          - kubewarden
      endpoints:
        - port: metrics
          interval: 10s
    - name: kubewarden-controller
      selector:
        matchLabels:
          app.kubernetes.io/name: kubewarden-controller
      namespaceSelector:
        matchNames:
          - kubewarden
      endpoints:
        - port: metrics
          interval: 10s
EOF

Let's install the Prometheus stack Helm Chart:

note

At time of writing the latest chart version is 51.5.3

helm repo add prometheus-community https://prometheus-community.github.io/helm-charts

helm install --wait --create-namespace \
  --namespace prometheus \
  --version 51.5.3 \
  --values kube-prometheus-stack-values.yaml \
  prometheus prometheus-community/kube-prometheus-stack

Install Kubewarden

We can now install Kubewarden in the recommended way with Helm charts.

note

cert-manager is a requirement of OpenTelemetry, but we've already installed them in a previous section of this book.

As a first step, we have to add the Helm repository that contains Kubewarden:

helm repo add kubewarden https://charts.kubewarden.io

Then we have to install the Custom Resource Definitions (CRDs) defined by Kubewarden:

helm install --wait \
  --namespace kubewarden --create-namespace \
  kubewarden-crds kubewarden/kubewarden-crds

Now we can deploy the rest of the Kubewarden stack. The official helm chart will create a PolicyServer named default.

Let's configure the values of the Helm Chart so that we have metrics enabled in Kubewarden. Write the kubewarden-values.yaml file with the following contents:

telemetry:
  mode: sidecar
  metrics: True
  sidecar:
    metrics:
      port: 8080

Now, let's install the helm charts:

helm install --wait \
  --namespace kubewarden \
  --create-namespace \
  --values kubewarden-values.yaml \
  kubewarden-controller kubewarden/kubewarden-controller

helm install --wait \
  --namespace kubewarden \
  --create-namespace \
  kubewarden-defaults kubewarden/kubewarden-defaults \
  --set recommendedPolicies.enabled=True \
  --set recommendedPolicies.defaultPolicyMode=monitor

This leads to the creation of the default instance of PolicyServer:

kubectl get policyservers.policies.kubewarden.io
NAME      AGE
default   3m7s

By default, this policy server will have metrics enabled.

Accessing Prometheus

Prometheus exposes a very simple UI that we can use to inspect metrics exposed by different components within our Kubernetes cluster.

We can forward the Prometheus port so we can easily access it.

kubectl port-forward -n prometheus --address 0.0.0.0 svc/prometheus-operated 9090

Now, we can visit prometheus through port 9090 and perform a query, for example: kubewarden_policy_evaluations_total. We will see that the number of evaluations will grow over time as we produce more requests that go through the policy.

Accessing Grafana

We can forward the Grafana service so we can easily access it.

kubectl port-forward -n prometheus --address 0.0.0.0 svc/prometheus-grafana 8080:80

You can now login with the default username admin and password prom-operator.

Using Kubewarden Grafana dashboard

The Kubewarden developers made available a Grafana dashboard with some basic metrics that give an overview about how Kubewarden behaves inside of cluster. This dashboard is available in the GitHub releases of the Kubewarden policy-server repository as a JSON file or in the Grafana website.

To import the dashboard into your environment, you can download the JSON file from the Grafana website or from the repository:

curl https://github.com/kubewarden/policy-server/releases/latest/download/kubewarden-dashboard.json

Once you have the file in your machine you should access the Grafana dashboard and import it. Visit /dashboard/import in the Grafana dashboard and follow these steps:

1. Copy the JSON file contents and paste them into the Import via panel json box in the Grafana UI
2. Click the Load button
3. Choosing Prometheus as the source
4. Click the Import button

Another option is import it directly from the Grafana.com website. For this:

1. Copy the dashboard ID from the dashboard page,
2. Paste it in the Import via grafana.com field
3. Click the load button.
4. After importing the dashboard, define the Prometheus data source to use and finish the import process.

The Grafana dashboard has panes showing the state of all the policies managed by Kubewarden. Plus it has policy-specific panels.

Policy detailed metrics can be obtained by changing the value of the policy_name variable to match the name of the desired policy.

You should be able to see the dashboard similar to this: