How to Use Deployment Strategy in Kubernetes

In Kubernetes, you use the strategy field in a Deployment to control how updates are applied to your pods. The two main strategies are RollingUpdate for gradual updates and Recreate for replacing all pods at once.

Syntax

The strategy field in a Kubernetes Deployment defines how pods are updated during a deployment. It has two main types:

• RollingUpdate: Updates pods gradually to avoid downtime.
• Recreate: Deletes all old pods before creating new ones.

For RollingUpdate, you can specify maxUnavailable and maxSurge to control how many pods are updated at once.

yaml

strategy:
  type: RollingUpdate
  rollingUpdate:
    maxUnavailable: 1
    maxSurge: 1

Example

This example shows a Deployment using the RollingUpdate strategy. It updates pods one at a time, ensuring at least some pods are always running.

yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: example-deployment
spec:
  replicas: 3
  selector:
    matchLabels:
      app: example
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 1
  template:
    metadata:
      labels:
        app: example
    spec:
      containers:
      - name: example-container
        image: nginx:1.21.0

Output

deployment.apps/example-deployment created

Common Pitfalls

Common mistakes when using deployment strategies include:

• Setting maxUnavailable or maxSurge too high, causing downtime or resource overload.
• Using Recreate strategy without understanding it causes downtime by deleting all pods before starting new ones.
• Not specifying a strategy, which defaults to RollingUpdate but may not fit all use cases.

yaml

# Wrong way (causes downtime):
strategy:
  type: Recreate

# Right way (rolling update with controlled pods):
strategy:
  type: RollingUpdate
  rollingUpdate:
    maxUnavailable: 1
    maxSurge: 1

Quick Reference

Strategy	Description	Use Case
RollingUpdate	Updates pods gradually without downtime	Most common for production updates
Recreate	Deletes all pods before creating new ones	Simple apps where downtime is acceptable

Key Takeaways

Use the strategy field in Deployment to control pod update behavior.

RollingUpdate is best for zero-downtime updates with controlled pod replacement.

Recreate strategy causes downtime by deleting all pods before new ones start.

Configure maxUnavailable and maxSurge to balance availability and speed.

Always test deployment strategy settings in a staging environment first.