Skip to content

Scale a Deployment

When to Use This Guide

Use this guide when you need to increase or decrease the number of running pods for an application.

Scaling deployments is commonly required when:

  • application traffic increases
  • testing high availability
  • reducing resource usage
  • preparing for load testing

Quick Commands

kubectl scale deployment <deployment-name> --replicas=5
kubectl get deployment <deployment-name>
kubectl get pods

These commands allow you to quickly adjust the number of pods running in a deployment.

Common Scaling Tasks

Task Command
Increase replicas kubectl scale deployment web-app --replicas=5
Reduce replicas kubectl scale deployment web-app --replicas=2
Check replica status kubectl get deployment web-app
View running pods kubectl get pods

Step 1 — Check Current Deployment Status

List deployments in the namespace.

kubectl get deployments

Example: 

NAME      READY   UP-TO-DATE   AVAILABLE   AGE
web-app   2/2     2            2           10m

This indicates the deployment currently runs 2 pods.

Step 2 — Scale the Deployment

Increase the number of replicas. 

kubectl scale deployment web-app --replicas=5

Output: 

deployment.apps/web-app scaled

Kubernetes immediately begins creating additional pods.

Step 3 — Verify the New Pods

Check that new pods are starting. 

kubectl get pods

Example: 

web-app-7f8d9c6d4b-abc12
web-app-7f8d9c6d4b-def34
web-app-7f8d9c6d4b-ghi56
web-app-7f8d9c6d4b-jkl78
web-app-7f8d9c6d4b-mno90

Step 4 — Confirm Deployment Status

Verify that the deployment has reached the desired state. 

kubectl get deployment web-app

Example: 

NAME      READY   UP-TO-DATE   AVAILABLE
web-app   5/5     5            5

Automatic Scaling with HPA

For dynamic workloads, Kubernetes can automatically scale deployments using the Horizontal Pod Autoscaler (HPA).

Create an autoscaler: 

kubectl autoscale deployment web-app --cpu-percent=70 --min=2 --max=10

This tells Kubernetes to:

  • maintain CPU usage around 70%
  • run 2–10 pods depending on load

Verify Autoscaling

Check autoscaler status: 

kubectl get hpa

Example: 

NAME      REFERENCE            TARGETS   MINPODS   MAXPODS
web-app   Deployment/web-app   40%/70%   2         10

Common Issues

New Pods Remain Pending

Possible causes:

  • insufficient cluster resources
  • node scheduling constraints
  • missing persistent volumes

Investigate using: 

kubectl describe pod <pod-name>

Pods Crash After Scaling

This usually indicates an application issue rather than a scaling problem.

Check logs: 

kubectl logs <pod-name>

Quick Scaling Workflow

Most scaling tasks follow this pattern: 

kubectl get deployments
kubectl scale deployment <deployment-name> --replicas=5
kubectl get pods

Deployment Management

  • Roll Back a Deployment
  • Restart a Deployment

Troubleshooting

  • Debug a Crashing Pod
  • View Container Logs