Post Conversion Cleanup: Cluster Autoscaler Configuration

Follow the steps in this page to ensure that the Cluster Autoscaler is able to function properly after converting your DKP Essential cluster to a DKP Enterprise Managed cluster.

After converting your cluster from Essential to Enterprise, the Cluster Lifecycle Management responsibilities is moved to a single Management cluster.

The Cluster Autoscaler feature also depends on the same Cluster Lifecycle Management components. If you are using the Cluster Autoscaler feature in DKP, you must perform the following steps for this feature to continue to work correctly:

Run the following commands in the Management cluster. For general guidelines on how to set the context, refer to Provide Context for Commands with a kubeconfig File.

Set the following environment variables with your cluster's details:
```
export CLUSTER_NAME=<>
export WORKSPACE_NAMESPACE=<>
```
CODE

Apply the Cluster Autoscaler Deployment and supporting resources:

cat <<EOF | kubectl apply -f -
---
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: cluster-autoscaler-${CLUSTER_NAME}
  name: cluster-autoscaler-${CLUSTER_NAME}
  namespace: ${WORKSPACE_NAMESPACE}
spec:
  replicas: 1
  selector:
    matchLabels:
      app: cluster-autoscaler-${CLUSTER_NAME}
  template:
    metadata:
      labels:
        app: cluster-autoscaler-${CLUSTER_NAME}
    spec:
      containers:
      - args:
        - --cloud-provider=clusterapi
        - --node-group-auto-discovery=clusterapi:clusterName=${CLUSTER_NAME}
        - --kubeconfig=/workload-cluster/kubeconfig
        - --clusterapi-cloud-config-authoritative
        - -v5
        command:
        - /cluster-autoscaler
        image: us.gcr.io/k8s-artifacts-prod/autoscaling/cluster-autoscaler:v1.25.0
        name: cluster-autoscaler
        volumeMounts:
        - mountPath: /workload-cluster
          name: kubeconfig
          readOnly: true
      serviceAccountName: cluster-autoscaler-${CLUSTER_NAME}
      terminationGracePeriodSeconds: 10
      tolerations:
      - effect: NoSchedule
        key: node-role.kubernetes.io/master
      - effect: NoSchedule
        key: node-role.kubernetes.io/control-plane
      volumes:
      - name: kubeconfig
        secret:
          items:
          - key: value
            path: kubeconfig
          secretName: ${CLUSTER_NAME}-kubeconfig
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: cluster-autoscaler-management-${CLUSTER_NAME}
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: cluster-autoscaler-management-${CLUSTER_NAME}
subjects:
- kind: ServiceAccount
  name: cluster-autoscaler-${CLUSTER_NAME}
  namespace: ${WORKSPACE_NAMESPACE}
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: cluster-autoscaler-${CLUSTER_NAME}
  namespace: ${WORKSPACE_NAMESPACE}
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: cluster-autoscaler-management-${CLUSTER_NAME}
rules:
- apiGroups:
  - cluster.x-k8s.io
  resources:
  - machinedeployments
  - machinedeployments/scale
  - machines
  - machinesets
  verbs:
  - get
  - list
  - update
  - watch
EOF

CODE

Verify the output is similar to the following:

deployment.apps/cluster-autoscaler-<cluster-name> created
clusterrolebinding.rbac.authorization.k8s.io/cluster-autoscaler-management-<cluster-name> created
serviceaccount/cluster-autoscaler-<cluster-name> created
clusterrole.rbac.authorization.k8s.io/cluster-autoscaler-management-<cluster-name> created

CODE

To check that the status of the deployment has the expected AVAILABLE count of 1, run the following command and verify that the output is similar:

$ kubectl get deployment -n $WORKSPACE_NAMESPACE cluster-autoscaler-$CLUSTER_NAME
NAME                                READY   UP-TO-DATE   AVAILABLE   AGE
cluster-autoscaler-<cluster-name>   1/1     1            1           1m

CODE

Next Step

Post Conversion Cleanup: Clusters run on Different Cloud Platforms

Next Step

See Also