Monitoring Kubernetes Node & Pod Resource Usage with Metrics Server

This guide explains how to deploy the Kubernetes Metrics Server in an Amazon EKS cluster and use it to monitor CPU and memory usage for nodes and pods.

📌 Overview

The metrics-server collects resource metrics from each node’s kubelet and exposes them through the Kubernetes Metrics API.
These metrics can then be viewed using commands like:

kubectl top nodes
kubectl top pods

1️⃣ Prerequisites

• A running EKS cluster
• kubectl installed and configured
• Permissions to create resources in the kube-system namespace

Check your current context:

kubectl config current-context

2️⃣ Deploy Metrics Server

Apply the official metrics-server manifest:

kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

This will create:

• metrics-server deployment in kube-system namespace
• Required ClusterRoles, RoleBindings, and APIService

3️⃣ EKS Compatibility Fix

On EKS, kubelet uses self-signed certificates, which can cause TLS errors in metrics-server. To fix this, add --kubelet-insecure-tls to the metrics-server container args.

Edit the deployment:

kubectl -n kube-system edit deployment metrics-server

Find the container args section and update it:

spec:
  template:
    spec:
      containers:
      - name: metrics-server
        args:
        - --cert-dir=/tmp
        - --secure-port=4443
        - --kubelet-preferred-address-types=InternalIP,Hostname,InternalDNS,ExternalDNS,ExternalIP
        - --kubelet-insecure-tls

Save and exit. The deployment will restart automatically.

4️⃣ Verify Metrics Server

Check if the deployment is running:

kubectl get deployment metrics-server -n kube-system

Expected output:

NAME             READY   UP-TO-DATE   AVAILABLE   AGE
metrics-server   1/1     1            1           1m

5️⃣ Viewing Node & Pod Usage

Once running, you can check node usage:

kubectl top nodes

Example output:

NAME                                          CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
ip-10-0-1-45.ap-south-2.compute.internal      450m         22%    1200Mi          60%
ip-10-0-2-30.ap-south-2.compute.internal      300m         15%    900Mi           45%

Check pod usage (all namespaces):

kubectl top pods -A

Example output:

NAMESPACE     NAME                          CPU(cores)   MEMORY(bytes)
default       nginx-5d59d67564-b8k2h         2m           6Mi
kube-system   coredns-66bff467f8-bn2fk       3m           12Mi

6️⃣ Filtering by Node Group or Labels

If your nodes are labeled (e.g., workload=emqx), you can filter:

kubectl top nodes -l workload=emqx

To view usage for only pods with a label:

kubectl top pods -A -l app=nginx

7️⃣ Common Issues

❌ error: Metrics API not available

• The metrics-server is not installed or not running.

• Fix: Check with:

  kubectl get deployment metrics-server -n kube-system

❌ x509: certificate signed by unknown authority

• Fix: Add --kubelet-insecure-tls in the deployment args.

8️⃣ Useful Commands Summary

9️⃣ Version Compatibility

For EKS with Kubernetes 1.33.x, use the latest v0.7.x release.

🔗 References

• Metrics Server GitHub (opens in a new tab)
• EKS Monitoring Docs (opens in a new tab)