Back to lesson: Choose the Right Workload Controller
Free lab · CKA Prep Program · Module 3: Workloads and scheduling
Rollout and scale
30 min hands-on · Core
Browser Kubernetes (Killercoda 1-node) or Play-with-Kubernetes; no local installs.
Two ways to do this lab: in your browser on Killercoda (free, no install), or on Play-with-Kubernetes as a free practice field. The Killercoda button opens in a new tab; this study page stays here. Killercoda runs one free scenario at a time: if you see a waiting queue, close other Killercoda tabs and wait a minute.
Objectives
- Deploy, scale and expose a workload imperatively
- Run a rolling update and watch it converge
- Roll back a bad rollout from history
Step 1
Deploy and expose
Create a deployment (`kubectl create deployment web --image=nginx:1.25`), scale it to 3, expose it ClusterIP. Verify with `kubectl get deploy,rs,pods,svc -o wide`.
Step 2
Rolling update
Change the image with `kubectl set image`, then watch `kubectl rollout status`. Record how many pods stay available during the roll; that number is the point of the exercise.
Step 3
Break it, roll it back
Set a nonexistent image tag, confirm the rollout stalls, then `kubectl rollout undo`. Verify the old ReplicaSet serves again.
How to confirm it worked
- Deployment at 3 replicas behind a ClusterIP Service
- Rollout status observed converging, availability number recorded
- Bad rollout stalled on record, rollback verified serving