World K1: The Food Court

Want 3, See 2

You will learn: how Kubernetes heals itself.

Checked with Kubernetes 1.37 on 31 Oct 2026

The problem

You are here: cluster kind-dev namespace default

~/shop $ kubectl get nodesNAME                STATUS     ROLES           AGE     VERSIONdev-control-plane   Ready      control-plane   6m22s   v1.37.0dev-worker          NotReady   <none>          6m22s   v1.37.0dev-worker2         Ready      <none>          6m22s   v1.37.0

Your app’s server died at 3 a.m. Everyone was asleep. Who fixes it?

Watch

Want3written on the order for shop

See33 ready

Do All goodGaps fixed: 0

dev-workerOpen

shop·d6b6z
shop·zbn9j

dev-worker2Open

shop·gnjxm

You tell Kubernetes what you want: 3 copies, always. This is the .

Watch

You are here: cluster kind-dev namespace default

~/shop $ kubectl get deployment shopNAME   READY   UP-TO-DATE   AVAILABLE   AGEshop   3/3     3            3           22s

A checks again and again. Want 3. See 3. All good.

Predict

A whole stall closes, and one tray is lost. What happens next?

Watch

You are here: cluster kind-dev namespace default

~/shop $ kubectl delete pod shop-747bf9cf8-d6b6zpod "shop-747bf9cf8-d6b6z" deleted from default namespace~/shop $ kubectl get podsNAME                   READY   STATUS    RESTARTS   AGEshop-747bf9cf8-gnjxm   1/1     Running   0          24sshop-747bf9cf8-wrfvr   0/1     Running   0          2sshop-747bf9cf8-zbn9j   1/1     Running   0          24s

Want 3. See 2. Make 1 more. That is the whole trick.

Try it

Tap “delete” on any pod. Watch what Kubernetes does.

Want3written on the order for shop

See33 ready

Do All goodGaps fixed: 0

dev-workerOpen

shop·d6b6z
shop·zbn9j

dev-worker2Open

shop·gnjxm

Try it

Now change the order. Type kubectl scale deployment shop --replicas=5.

You are here: cluster kind-dev namespace default

Want3written on the order for shop

See33 ready

Do All goodGaps fixed: 0

dev-control-planeOpenManager

No trays here.

dev-workerOpen

shop·d6b6z
shop·zbn9j

dev-worker2Open

shop·gnjxm

Time: +0s (sped up)

Check

Tap the new pod.

0 of 1 found

The new pod has a new name. Kubernetes replaces pods; it does not repair them.

You did it!

You can now explain how Kubernetes heals itself.

Go deeper (optional)

This loop is called . If only a container crashes, the kubelet restarts it in the same pod. Watch the RESTARTS column. When a node dies, Kubernetes waits about 5 minutes. Then it replaces the pods.

You are here: cluster kind-dev namespace default

~/shop $ kubectl delete pod shop-747bf9cf8-d6b6z --wait=falsepod "shop-747bf9cf8-d6b6z" deleted from default namespace~/shop $ kubectl get pods -wNAME                   READY   STATUS        RESTARTS   AGEshop-747bf9cf8-d6b6z   1/1     Terminating   0          22sshop-747bf9cf8-gnjxm   1/1     Running       0          22sshop-747bf9cf8-wrfvr   0/1     Pending       0          0sshop-747bf9cf8-zbn9j   1/1     Running       0          22sshop-747bf9cf8-wrfvr   0/1     Running       0          1sshop-747bf9cf8-d6b6z   1/1     Terminating   0          23sshop-747bf9cf8-wrfvr   1/1     Running       0          6s^C

Real mission

On your own kind cluster, order 3 trays.
kubectl create deployment shop --image=nginx:1.29-alpine --replicas=3
Look at the trays and their names.
kubectl get pods
Delete one. Copy a real name from the list first.
kubectl delete pod <a shop pod name>
Look again: 3 trays, one with a new name.
kubectl get pods

0 of 4 steps done · Your ticks are saved.