Skip to content

Cluster Proportional Autoscaler

Horizontal cluster-proportional-autoscaler watches over the number of schedulable nodes and cores of the cluster and resizes the number of replicas for the required resource. This functionality may be desirable for applications that need to be autoscaled with the size of the cluster, such as CoreDNS and other services that scale with the number of nodes/pods in the cluster.

The cluster-proportional-autoscaler helps to scale the applications using deployment or replicationcontroller or replicaset. This is an alternative solution to Horizontal Pod Autoscaling. It is typically installed as a Deployment in your cluster.

Refer to the eks-best-practices-guides for addional configuration guidanance.

Usage

This add-on requires both enable_cluster_proportional_autoscaler and cluster_proportional_autoscaler as mandatory fields.

The example shows how to enable cluster-proportional-autoscaler for CoreDNS Deployment. CoreDNS deployment is not configured with HPA. So, this add-on helps to scale CoreDNS Add-on according to the size of the nodes and cores.

This Add-on can be used to scale any application with Deployment objects.

enable_cluster_proportional_autoscaler  = true
cluster_proportional_autoscaler  = {
    values = [
      <<-EOT
        nameOverride: kube-dns-autoscaler

        # Formula for controlling the replicas. Adjust according to your needs
        # replicas = max( ceil( cores * 1/coresPerReplica ) , ceil( nodes * 1/nodesPerReplica ) )
        config:
          linear:
            coresPerReplica: 256
            nodesPerReplica: 16
            min: 1
            max: 100
            preventSinglePointFailure: true
            includeUnschedulableNodes: true

        # Target to scale. In format: deployment/*, replicationcontroller/* or replicaset/* (not case sensitive).
        options:
          target: deployment/coredns # Notice the target as `deployment/coredns`

        serviceAccount:
          create: true
          name: kube-dns-autoscaler

        podSecurityContext:
          seccompProfile:
            type: RuntimeDefault
            supplementalGroups: [65534]
            fsGroup: 65534

        resources:
          limits:
            cpu: 100m
            memory: 128Mi
          requests:
            cpu: 100m
            memory: 128Mi

        tolerations:
          - key: "CriticalAddonsOnly"
            operator: "Exists"
            description: "Cluster Proportional Autoscaler for CoreDNS Service"
      EOT
    ]
}

Expected result

The cluster-proportional-autoscaler pod running in the kube-system namespace. 

kubectl -n kube-system get po -l app.kubernetes.io/instance=cluster-proportional-autoscaler
NAME                                                              READY   STATUS    RESTARTS   AGE
cluster-proportional-autoscaler-kube-dns-autoscaler-d8dc8477xx7   1/1     Running   0          21h
The cluster-proportional-autoscaler-kube-dns-autoscaler config map exists. 
kubectl -n kube-system get cm cluster-proportional-autoscaler-kube-dns-autoscaler
NAME                                                  DATA   AGE
cluster-proportional-autoscaler-kube-dns-autoscaler   1      21h

Testing

To test that coredns pods scale, first take a baseline of how many nodes the cluster has and how many coredns pods are running. 

kubectl get nodes
NAME                          STATUS   ROLES    AGE   VERSION
ip-10-0-19-243.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-25-182.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-40-138.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-8-136.ec2.internal    Ready    <none>   21h   v1.26.4-eks-0a21954

kubectl get po -n kube-system -l k8s-app=kube-dns
NAME                       READY   STATUS    RESTARTS   AGE
coredns-7975d6fb9b-dlkdd   1/1     Running   0          21h
coredns-7975d6fb9b-xqqwp   1/1     Running   0          21h

Change the following parameters in the hcl code above so a scaling event can be easily triggered: 

        config:
          linear:
            coresPerReplica: 4
            nodesPerReplica: 2
            min: 1
            max: 4
and execute terraform apply.

Increase the managed node group desired size, in this example from 4 to 5. This can be done via the AWS Console.

Check that the new node came up and coredns scaled up. 

NAME                          STATUS   ROLES    AGE   VERSION
ip-10-0-14-120.ec2.internal   Ready    <none>   10m   v1.26.4-eks-0a21954
ip-10-0-19-243.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-25-182.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-40-138.ec2.internal   Ready    <none>   21h   v1.26.4-eks-0a21954
ip-10-0-8-136.ec2.internal    Ready    <none>   21h   v1.26.4-eks-0a21954

kubectl get po -n kube-system -l k8s-app=kube-dns
NAME                       READY   STATUS    RESTARTS   AGE
coredns-7975d6fb9b-dlkdd   1/1     Running   0          21h
coredns-7975d6fb9b-ww64t   1/1     Running   0          10m
coredns-7975d6fb9b-xqqwp   1/1     Running   0          21h