Test Cluster Autoscaler

Refer to these instructions to test an installed cluster autoscaler.

Requirements

This task assumes you have installed the cluster autoscaler on a TKG cluster.

Test Cluster Autoscaler

To verify that the autoscaler automatically scales worker nodes, deploy an application and then scale the number of replicas in the deployment. The autoscaler will scale up worker nodes once the node resources are insufficient.

Create the following application definition named app.yaml.

apiVersion: v1
kind: Namespace
metadata:
  name: app
  labels:
    pod-security.kubernetes.io/enforce: privileged
---
apiVersion: v1
kind: Service
metadata:
  name: application-cpu
  namespace: app
  labels:
    app: application-cpu
spec:
  type: ClusterIP
  selector:
    app: application-cpu
  ports:
    - protocol: TCP
      name: http
      port: 80
      targetPort: 80
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: application-cpu
  namespace: app
  labels:
    app: application-cpu
spec:
  selector:
    matchLabels:
      app: application-cpu
    replicas: 1
    strategy:
      type: RollingUpdate
      rollingUpdate:
        maxSurge: 1
        maxUnavailable: 0
    template:
      metadata:
        labels:
          app: application-cpu
      spec:
        containers:
        - name: application-cpu
          image: wcp-docker-ci.artifactory.eng.vmware.com/app-cpu:v1.0.0
          imagePullPolicy: Always
          ports:
            -containerPort: 80
          resources:
            requests:
              memory: 50Mi
              cpu: 500m
            limits:
              memory: 500Mi
              cpu: 2000m

Create the application.
```
kubectl apply -f app.yaml
```
Scale up the application's replicas to trigger the autoscaler.
For example, increase the number of spec.selector.replicas from 1 to a larger number such that additional worker nodes are required.
Update the application.
```
kubectl apply -f app.yaml
```
Verify that additional worker nodes are created to handle the load.
The autoscaler will scale up the number of worker nodes once the node resources are insufficient.