Deploy metrics server in kubernetes for auto horizontal scaling

1. Get the metrics server code form github

git clone https://github.com/kubernetes-sigs/metrics-server
cd metrics-server

#Edit metrics-server-deployment.yaml
vi deploy/kubernetes/metrics-server-deployment.yaml

#And add below args

args:
 - --kubelet-preferred-address-types=InternalIP,Hostname,InternalDNS,ExternalDNS,ExternalIP
 - --kubelet-insecure-tls

metrics-server-deployment.yaml will look like below

2. After deployment we will get the cpu and ram usage of node as below

3.Now we can write Horizontal Pod Autoscaler as below that will auto scale nginx-app1 deplyment if cpu usage will get above 80% max 5 pods.
– It’s checks every 30 seconds for scaling the deployment
– It’s scale downs the deployment after 300 seconds if the load goes down

kind: HorizontalPodAutoscaler
apiVersion: autoscaling/v1
metadata:
  name: nginx-app1-hpa
spec:
  scaleTargetRef:
    kind: Deployment
    name: nginx-app1
    apiVersion: apps/v1
  minReplicas: 1
  maxReplicas: 5
  targetCPUUtilizationPercentage: 80

4. nginx-app1.yml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-app1
spec:
  selector:
    matchLabels:
      run: nginx-app1
  replicas: 2
  template:
    metadata:
      labels:
        run: nginx-app1
    spec:
      containers:
      - name: nginx-app1
        image: nginx
        resources:
          requests:
            memory: "128Mi"
            cpu: "100m"
          limits:
            memory: "256Mi"
            cpu: "200m"
        ports:
        - containerPort: 80

---

kind: Service
apiVersion: v1
metadata:
  name: nginx-app1-svc
  labels:
    run: nginx-app1-svc
spec:
  ports:
    - protocol: TCP
      port: 80
      targetPort: 80
      nodePort: 30083
  selector:
    run: nginx-app1
  type: NodePort

5. Random load generator

while(true)
do
curl -s http://SERVICE_NAME
curl -s http://SERVICE_NAME
curl -s http://SERVICE_NAME
curl -s http://SERVICE_NAME
curl -s http://SERVICE_NAME
done