Skip to main content
In this lesson you’ll deploy the Kubernetes Metrics Server and use kubectl top to monitor CPU and memory usage for nodes and pods. The lab environment already runs three application pods: elephant, lion, and rabbit.

​
Verify running pods

Start by listing the pods in the cluster. You should see the three workload pods in Running state:

​
Clone the prepared Metrics Server repository

This lab supplies a test repository with manifests tailored for the exercise. Clone it to the control plane and inspect the files:
Do not use the lab repository directly in production. The manifests may be simplified or modified for the lab environment. For production deployments, follow the official Metrics Server documentation: https://github.com/kubernetes-sigs/metrics-server
A dark-themed GitHub repository page for "kubernetes-metrics-server" showing a list of YAML files (e.g., metrics-server-deployment.yaml) and the README content. The right sidebar displays repo stats (stars, forks) and contributor avatars.

​
Deploy the Metrics Server manifests

From the cloned repository directory, create the required resources to install the Metrics Server:
The Metrics Server may take a short while to become ready and start publishing metrics to the Metrics API. Wait a minute, then use kubectl top to query node and pod metrics.

​
Use kubectl top to view resource usage

Once the Metrics Server is collecting metrics, kubectl top reports CPU and memory usage for nodes and pods. Example output for nodes:
Example output for pods:

​
Interpretation — identify heavy resource consumers

Use the kubectl top results to answer common monitoring questions. The primary consumers in this example are: And the direct answers to the lab questions: That completes the lab: you deployed the Metrics Server and used kubectl top to identify nodes and pods that consume the most resources.

Watch Video