How to send metrics from your Kubernetes cluster to your Cockpit

In this page we will show you how to send application metrics from your Kubernetes cluster to your Cockpit, either by using a Helm chart or deploying a Helm chart with Terraform.

We will use the k8s-monitoring Helm Chart, which installs an Alloy Daemon set to export your Kubernetes cluster’s metrics to your Cockpit.

Before you start

To complete the actions presented below, you must have:

• A Scaleway account metricsged into the console
• Owner status or IAM permissions allowing you to perform actions in the intended Organization
• Created a custom external data source of the metrics type
• Created a Cockpit token in the same region as the metrics data source
• A running Kubernetes cluster containing your deployed application
• Created an API key and retrieved your API secret key

Configure the Helm chart

Create a values.yml file to configure your Helm chart, using the example below. Make sure that you replace $SCW_CLUSTER_NAME with the name of your Scaleway Kubernetes cluster, $COCKPIT_CUSTOM_METRICS_DATASOURCE_URL with the URL of your custom metrics data source (you can find it under the “API URL” section in the Data sources tab of the Scaleway console), and $COCKPIT_TOKEN with your Cockpit token.

cluster:
  name: "$SCW_CLUSTER_NAME"
global:
  scrape_interval: 60s
destinations:
  - name: "my-cockpit-metrics"
    type: "prometheus"
    protocol: "http"
    metrics:
      enabled: true
    url: "$COCKPIT_CUSTOM_METRICS_DATASOURCE_URL/api/v1/push"
    tenantId: "$COCKPIT_TOKEN"
    logs:
      enabled: false
    traces:
      enabled: false
annotationAutodiscovery:
  enabled: true
  destinations: ["my-cockpit-metrics"]
alloy-metrics:
  enabled: true
alloy-singleton:
  enabled: true

Add annotations for auto-discovery

Annotations in Kubernetes provide a way to attach metadata to your resources. For k8s-monitoring, these annotations signal which pods should be scraped for metrics, and what port to use. For the sake of this documentation, we are adding annotations to specify we want k8s-monitoring to scrape the pods from our deployment. Make sure that you replace $METRICS_PORT with the port where your application exposes Prometheus metrics.

Kubernetes deployment template

apiVersion: apps/v1
kind: Deployment
metadata:
  ...
  annotations:
    "k8s.grafana.com/metrics.portNumber" = "$METRICS_PORT"
    "k8s.grafana.com/scrape" = "true"
spec:
  ...

Terraform deployment template

resource "kubernetes_deployment_v1" "your_application_deployment" {
  ...
  spec {
    ...
    template {
      metadata {
        ...
        annotations = {
          "k8s.grafana.com/metrics.portNumber" = "$METRICS_PORT"
          "k8s.grafana.com/scrape" = "true"
        }
      }
      ...
    }
  }
}

Send Kubernetes metrics using Helm chart with Terraform

1. Create a provider.tf file and paste the following template to set up the Helm Terraform provider:

   provider "helm" {
     kubernetes {
       host = your_k8s_cluster_host # The URL of your Kubernetes API server.
       token = your_k8s_cluster_token # Authentication token to access the cluster.
       cluster_ca_certificate = base64decode(
       your_k8s_cluster_ca_certificate # The cluster's CA certificate.
       )
     }
   }
   

2. Create a maint.tf file and paste the following template to create a Helm release resource. Make sure that you replace /your-path/to/values.yml with the actual path to your values file.

   resource "helm_release" "alloy" {
     name = "name-of-your-metrics-ingester"
     repository = "https://grafana.github.io/helm-charts"
     chart = "k8s-monitoring"
     version = "2.0.2"
     namespace = "metrics-ingester"
     create_namespace = true
     values = [file("/your-path/to/values.yml")]
   }
   

3. Save your changes.
4. Run terraform init to initialize your Terraform configuration and download any necessary providers.
5. Run terraform apply to apply your configuration.
6. Type yes when prompted to confirm the actions.

Send Kubernetes metrics using Helm chart

Once you have configured your values.yml file, you can use Helm to deploy the metric-forwarding configuration to your Kubernetes cluster. Before installing the Helm chart, ensure that your kubectl tool is properly connected to your Kubernetes cluster. kubectl is the command-line tool for interacting with Kubernetes clusters.

1. Connect kubectl to your Kubernetes cluster

2. Run the command below to install the k8s-monitoring Helm chart:

   helm install -f /your-path/to/values.yml name-of-your-choice-for-your-metric-ingester k8s-monitoring --version 1.6.16
   

   The -f flag specifies the path to your values.yml file, which contains the configuration for the Helm chart. Make sure that you replace /your-path/to/values.yml with the correct path where your values.yml file is stored. Make sure that you also replace name-of-your-choice-for-your-metric-ingester with a clear name (ex. alloy-metrics-ingester). In our configuration, we are using alloy-lm-ingester. Helm installs the k8s-monitoring chart, which includes the Alloy DaemonSet configured to collect metrics from your Kubernetes cluster. The DaemonSet ensures that a pod is running on each node in your cluster, which collects metrics and forwards them to the specified Prometheus endpoint in your Cockpit.

3. Optionally, check the status of the release to ensure it was installed:

   helm list
   

Explore your metrics in Cockpit

Now that your metrics are exported to your Cockpit, you can access and query them.

1. Click Cockpit in the Observability section of the Scaleway console side menu. The Cockpit Overview page displays.
2. Click Open dashboards to open your managed dashboards in Grafana. You are redirected to the Grafana website.
3. Log in to Grafana using your Grafana credentials.
4. Click the Home icon, then click Explore.
5. Select your custom data source in the search drop-down on the upper left corner of your screen.
6. In the Labels filter drop-down, select the cluster label and in the Value drop-down, select your cluster.
7. Optionally, click the Clock icon on the top right corner of your screen and filter by time range.
8. Click Run query to see your metrics. An output similar to the following should display.