Using HorizontalPodAutoscaler in Kubernetes

Introduction

HorizontalPodAutoscaler is a Kubernetes feature that allows you to automatically scale the number of pods in a deployment based on resource utilization. In this lab, we will learn how to use HorizontalPodAutoscaler to automatically scale our deployment.

Start the Minikube Cluster

Before creating resources, you need a running Kubernetes cluster. Minikube is a lightweight Kubernetes environment that runs on your local machine.

1. Navigate to your working directory:

   Open the terminal and navigate to the default project folder:

   cd /home/labex/project
   

2. Start Minikube:

   Start Minikube to initialize a Kubernetes cluster:

   minikube start
   

   • This command sets up a single-node Kubernetes cluster on your local machine.
   • Minikube may take a few minutes to start depending on your system's performance.

3. Verify Minikube is running:

   Check the status of the Minikube cluster:

   minikube status
   

   • Look for components like kubelet and apiserver listed as Running.
   • If the cluster is not running, rerun minikube start.

If you encounter issues starting Minikube. Use minikube delete to reset the environment if needed.

Create a Deployment

First, we need to create a deployment to which we will apply the HorizontalPodAutoscaler.

1. Create a deployment file named deployment.yaml with the following contents:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: hpa-demo
spec:
  replicas: 1
  selector:
    matchLabels:
      app: hpa-demo
  template:
    metadata:
      labels:
        app: hpa-demo
    spec:
      containers:
        - name: hpa-demo
          image: nginx
          resources:
            limits:
              cpu: "1"
              memory: 512Mi
            requests:
              cpu: "0.5"
              memory: 256Mi
---
apiVersion: v1
kind: Service
metadata:
  name: hpa-demo
spec:
  selector:
    app: hpa-demo
  ports:
    - name: http
      port: 80
      targetPort: 80

This deployment specifies a single replica of an Nginx container with resource limits and requests for CPU and memory.

2. Create the deployment:

kubectl apply -f deployment.yaml

Create a HorizontalPodAutoscaler

Now that we have a deployment, we can create a HorizontalPodAutoscaler to automatically scale the deployment.

1. Create a HorizontalPodAutoscaler file named hpa.yaml with the following contents:

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: hpa-demo
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: hpa-demo
  minReplicas: 1
  maxReplicas: 10
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          averageUtilization: 1
          type: Utilization

This HorizontalPodAutoscaler specifies that we want to scale the hpa-demo deployment to have between 1 and 10 replicas, and that we want to target an average CPU utilization of 50%.

2. Create the HorizontalPodAutoscaler:

kubectl apply -f hpa.yaml

Test the HorizontalPodAutoscaler

Now that we have a HorizontalPodAutoscaler, we can test it by generating load on the deployment.

1. Enable the metrice-server

minikube addons enable metrics-server

2. Create a load generation pod:

kubectl run -i --tty load-generator --image=busybox /bin/sh

3. In the load generation pod, run the following command to generate load on the deployment:

while true; do wget -q -O- http://hpa-demo; done

4. Open another terminal, check the status of the HorizontalPodAutoscaler:

kubectl get hpa

You can see that the number of copies of hpa-demo has been extended to 10. You can check the number of replicas with the following command.

kubectl get pods -l app=hpa-demo

5. Stop the load generation by typing ctrl+c in the load generation pod.

Summary

In this lab, we learned how to use HorizontalPodAutoscaler to automatically scale a deployment based on resource utilization. We created a deployment, created a HorizontalPodAutoscaler, and tested it by generating load on the deployment. We also saw how the HorizontalPodAutoscaler scaled the deployment in response to the increased load.