KubernetesIntermediate9 min read2026-03-01

Kubernetes Deployment: Rolling Updates and Rollbacks

Deploy scalable, zero-downtime microservices using Kubernetes Deployments, rolling updates, and rollbacks.

Prerequisites

Kubernetes basics

1. Writing a Declarative Deployment Manifest

Define desired replica counts, container specs, resource limits, and health probes.

yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: api-deployment
  labels:
    app: api
spec:
  replicas: 3
  selector:
    matchLabels:
      app: api
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
  template:
    metadata:
      labels:
        app: api
    spec:
      containers:
        - name: server
          image: nginx:1.25-alpine
          ports:
            - containerPort: 80
          resources:
            requests:
              cpu: "100m"
              memory: "128Mi"
            limits:
              cpu: "500m"
              memory: "512Mi"

2. Triggering Updates and Rollbacks

Update the container image and inspect rollout history.

Best Practices & Architecture Advice

Set maxUnavailable: 0 during rolling updates to guarantee zero downtime.
Always configure readiness probes to ensure traffic is only routed to fully initialized pods.

Common Mistakes to Watch Out For

•Using image: myapp:latest, which prevents Kubernetes from detecting that a new image version needs to be rolled out.

Frequently Asked Questions

How does Kubernetes roll back a failed deployment?

Run 'kubectl rollout undo deployment/<name>' to revert the replica set to the previous revision.