How to Monitor Amazon EKS Using Prometheus and Grafana (Without Helm)

Source: Dev.to

Why Monitoring Is Important

• Check cluster health
• Monitor node and pod resource usage
• Find issues early
• Improve system stability and security

Tools Used

• Amazon EKS
• Prometheus
• Grafana
• kube-state-metrics
• node-exporter

Helm is not used in this setup.

Simple Architecture

• Prometheus collects metrics from the EKS cluster
• Node Exporter collects node metrics
• kube-state-metrics provides Kubernetes object data
• Grafana displays metrics in dashboards

Step 1: Create Monitoring Namespace

(Create the namespace with your preferred kubectl command.)

Step 2: Deploy kube-state-metrics

Components deployed

• Pods
• Namespaces
• Nodes
• Deployments

Verify deployment

kubectl get pods -n monitoring

Step 3: Deploy Node Exporter

Collects

• CPU
• Memory
• Disk metrics

Check status with your usual kubectl commands.

Step 4: Configure Prometheus

Update Prometheus scrape configuration to collect metrics from:

• Kubernetes nodes
• Pods
• kube-state-metrics
• kubelet metrics

Relevant labels for Grafana filtering

• cluster
• node
• namespace
• pod

Step 5: Connect Grafana to Prometheus

1. Open Grafana
2. Add Prometheus as a data source
3. Save and test the connection

Step 6: Import Dashboards

Import Kubernetes dashboards to view:

• Cluster metrics
• Node usage
• Pod usage

You can filter by:

• Cluster
• Node
• Namespace
• Pod

What You Can Monitor

• Cluster health
• Node CPU and memory
• Pod resource usage
• Pod restarts

What I Learned

• Monitoring is essential for Kubernetes
• Prometheus and Grafana work well without Helm
• Labels are very useful for filtering metrics
• Good monitoring supports DevSecOps practices

Conclusion

This is a simple and effective way to monitor Amazon EKS without Helm. It’s a great starting point for anyone learning Kubernetes monitoring.

Thanks for reading!