Example Documentation: Rolling Update Failure
Goal
Observe a broken rolling update and recover it with a rollback.
What Is Done
First, a working Deployment is rolled out. Then a full update manifest with an intentionally broken readiness probe is applied.
Typical Commands
kubectl apply -f manifests/00-working.yaml
kubectl rollout status deployment/rollout-demo -n rollout-lab
kubectl apply -f manifests/90-broken-update.yaml
kubectl rollout history deployment/rollout-demo -n rollout-lab
kubectl rollout undo deployment/rollout-demo -n rollout-lab
Validate Success
- The first version becomes ready.
- The broken update remains stuck during rollout.
rollout undorestores the previous version.
Common Mistakes
rollout statusis run without a timeout and blocks for a long time.- The broken probe is mistaken for an image issue.
revisionHistoryLimitis too low for useful rollback exercises.