Skip to main content

ReplicaSet: Why You Rarely Create It Directly

Hands-on example​

Assume:

  • A working Kubernetes cluster
  • kubectl is configured

Initial state​

No ReplicaSets or Pods exist.

kubectl get rs
kubectl get pods
No resources found in default namespace.
No resources found in default namespace.

Step 1: Create a ReplicaSet directly​

Create a ReplicaSet using kubectl create.

kubectl create rs demo-rs --image=nginx --replicas=1
replicaset.apps/demo-rs created

Check resources.

kubectl get rs
kubectl get pods
NAME      DESIRED   CURRENT   READY   AGE
demo-rs   1         1         1       5s
NAME            READY   STATUS    RESTARTS   AGE
demo-rs-abcde   1/1     Running   0          5s

What changed:

  • One ReplicaSet exists
  • One Pod was created by the ReplicaSet

What did not change:

  • No Deployment exists

Step 2: Delete the Pod​

kubectl delete pod demo-rs-abcde
pod "demo-rs-abcde" deleted

Check Pods again.

kubectl get pods
NAME            READY   STATUS    RESTARTS   AGE
demo-rs-fghij   1/1     Running   0          5s

What changed:

  • The original Pod was deleted
  • A new Pod was created automatically

What did not change:

  • The ReplicaSet configuration stayed the same

Key observation​

  • ReplicaSets self-heal Pods but cannot do rolling updates
  • Use Deployments instead of creating ReplicaSets directly