Loading auth...
Medium10 minContainer Orchestration
UpdatedAug 2, 2026
Edit

RollingUpdate vs. Recreate Strategy

CONCEPTS:Deployment Strategies (RollingUpdate vs. Recreate)Kubernetes Health Probes

Why This Is Asked

Managing updates is a daily task for DevOps engineers. This question tests whether a candidate understands the trade-offs between zero-downtime (RollingUpdate) and consistency (Recreate), and how to configure them to prevent service interruptions.

Key Concepts

  • Downtime: Which strategy causes it and why.
  • Version Skew: The challenge of running two versions of an app at once.
  • MaxSurge vs. MaxUnavailable: Fine-tuning the update speed and availability.
  • Rollbacks: How Kubernetes handles a failed update.

Answers by Technology

+ Add Variant
KubernetesImprove this answer ✏️

Expected Answer

Kubernetes provides two built-in ways to update your application:

  1. RollingUpdate (Default): This strategy replaces old Pods with new ones one-by-one (or in small batches). It provides zero downtime because some Pods are always available to serve traffic. However, you must ensure your application and database can handle two different versions running at the same time.
  2. Recreate: This strategy kills all old Pods first, and only then starts the new ones. This results in downtime, but it guarantees that you never have two different versions of your code running simultaneously.

Why It Matters

Most modern web applications use RollingUpdate to avoid interrupting users. However, if your application is a legacy singleton that cannot have two instances running (e.g., due to file locking or a non-backward-compatible DB migration), you might be forced to use Recreate.

Example Code

Configuring a RollingUpdate

spec:
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 25%       # Create 25% more pods during update
      maxUnavailable: 0   # Don't kill any old pods until new ones are ready

Configuring a Recreate Strategy

spec:
  strategy:
    type: Recreate

Common Mistakes

  • Setting maxUnavailable to 100%: This effectively turns a RollingUpdate into a Recreate strategy with downtime.
  • Forgetting Readiness Probes: If you don’t have readiness probes, Kubernetes will assume a new Pod is “ready” as soon as the process starts, killing the old Pod before the new one is actually able to handle requests.
  • Not considering Database Migrations: Rolling out a new version that requires a breaking DB change using RollingUpdate will crash the old Pods that are still running.

Follow-up Questions

  • How do you undo a failed deployment? (Answer: Using kubectl rollout undo deployment/<name>).
  • What is the difference between Blue/Green and Canary? (Answer: Blue/Green switches 100% of traffic at once after the new version is tested. Canary sends a tiny slice of traffic to the new version first to “test the waters” in production).
  • What is maxSurge? (Answer: The number of extra Pods that can be created above the desired replica count during an update).

References