🚀 Multi-Cluster Deployment with GitOps (EKS + Argo CD)

This guide walks through setting up GitOps using Argo CD across multiple EKS clusters (1 Hub + 2 spokes), and managing them using Argo CD installed on the hub cluster.

Step 1 : Create EKS cluster

eksctl create cluster --name hub-cluster --region ap-south-1
eksctl create cluster --name spoke-cluster-1 --region ap-south-1
eksctl create cluster --name spoke-cluster-2 --region ap-south-1

verify in console

Step 2: Set kubectl Context to Hub Cluster

Step 3: Install Argo CD in Hub Cluster

kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml

verify all pods are up and running

Step 4: Enable Insecure Mode for Argo CD

Edit the argocd-cm ConfigMap:

kubectl edit configmap argocd-cm -n argocd

Add:

server.insecure: "true"

Step 5: Change Argo CD Service Type to NodePort

Edit the service:

kubectl edit svc argocd-server -n argocd

Change:

type: ClusterIP

to:

type: NodePort

Note the NodePort (e.g., 32583).

Step 6: Update Security Group Inbound Rules

Add inbound rules:

• TCP 80 (HTTP)

• TCP 32583 (Argo CD NodePort)

You should now be able to access Argo CD UI at:
http://<hub-cluster-node-ip>:32583

Step 7: Get Initial Admin Password

kubectl -n argocd get secret argocd-initial-admin-secret -o jsonpath="{.data.password}" | base64 -d

🪟 On Windows PowerShell:

System.Text.Encoding]::UTF8.GetString([System.Convert]::FromBase64String("<BASE64_PASSWORD>"))

Step 8: Install Argo CD CLI

Download from:
👉 https://argo-cd.readthedocs.io/en/stable/cli_installation/

Step 9: Login Using Argo CD CLI

argocd login <hub-cluster-node-ip>:32583 --username admin --password <decoded-password> --insecure

Step 10: Register Additional Clusters (Spoke Clusters)

Switch context to each spoke cluster and register:

bashCopyEditaws eks --region ap-south-1 update-kubeconfig --name spoke-cluster-1
argocd cluster add <context-name> --server <hub-cluster-ip>:32583 --insecure

aws eks --region ap-south-1 update-kubeconfig --name spoke-cluster-2
argocd cluster add <context-name> --server <hub-cluster-ip>:32583 --insecure

🔍 Check in Argo CD UI under Settings > Clusters
Only in-cluster is shown by default — new clusters will appear after this step.

Step 11: Create an Application in Argo CD

You can create an Argo CD Application from:

• CLI

• YAML

• UI

Example CLI:

bashCopyEditargocd app create guestbook \
  --repo https://github.com/argoproj/argocd-example-apps.git \
  --path guestbook \
  --dest-server https://<spoke-cluster-api-endpoint> \
  --dest-namespace default \
  --server <hub-cluster-ip>:32583

tep 12: Test GitOps Behavior

Make a Git Commit (e.g., replicas: 3)

Push the change. Argo CD will reconcile and apply it.

Try Manual Override:

kubectl scale deployment guestbook-ui --replicas=1

⚠️ Replica count reverts back to 3.
This is because self-heal is enabled.

any manual changes you make (like scaling via kubectl) are quickly reverted by Argo CD to match the desired state defined in Git.

We'll update the replica count to 1 in the Git manifest. Once committed and pushed, Argo CD will pick up the change and reconcile the deployment by scaling down the replicas accordingly.

🧹 Cleanup

To delete all resources:

eksctl delete cluster --name hub-cluster --region ap-south-1
eksctl delete cluster --name spoke-cluster-1 --region ap-south-1
eksctl delete cluster --name spoke-cluster-2 --region ap-south-1

Also clean up:

• Git repositories

• EC2 security group rules

✅ Conclusion

In this hands-on guide, we successfully implemented a multi-cluster GitOps setup using Argo CD across Amazon EKS clusters. By deploying Argo CD on a hub cluster and connecting multiple spoke clusters, we enabled centralized management of Kubernetes workloads via Git as the single source of truth. The power of Argo CD's self-healing and declarative deployment was clearly demonstrated — any manual intervention was automatically reverted to match the Git state. This showcases the true potential of GitOps in production-grade environments.

This architecture lays a strong foundation for scalable, secure, and automated Kubernetes operations across multiple environments or teams.