Rollback and Revision History
What you'll learn: How to safely undo Deployment changes and manage the history of your rollouts.
Every time you update a Deployment (changing an image, labels, or pod template), Kubernetes creates a new revision—a snapshot of that version. This history acts like an undo stack, letting you revert to previous working configurations when something goes wrong.
How Rollback Works
When a new rollout causes problems (crashes, errors, performance issues), you can instantly roll back using:
kubectl rollout undo deployment/my-deployment
This command reverts to the previous revision by default. Think of it like hitting "Undo" in a text editor—Kubernetes restores the prior configuration and performs a rolling update backward.
Rolling Back to a Specific Revision
You can view your Deployment's history:
kubectl rollout history deployment/my-deployment
To jump back to a specific revision (say, revision 3):
kubectl rollout undo deployment/my-deployment --to-revision=3
Managing Revision History Limits
Kubernetes doesn't keep every revision forever. The revisionHistoryLimit field (default: 10) controls how many old ReplicaSets are retained. In your Deployment spec:
spec:
revisionHistoryLimit: 5
Setting this too low means fewer rollback options; too high consumes more cluster resources storing old ReplicaSets.
Analogy
Imagine revisions as checkpoints in a video game. Each major change creates a save point you can return to if things go wrong. revisionHistoryLimit determines how many save slots you keep—balance safety with storage space.
Key Takeaway: Use kubectl rollout undo to instantly revert failed Deployments, view history with rollout history, and control storage with revisionHistoryLimit to balance recovery options against cluster overhead.