Skip to main content

Autoscaling CoreDNS

CoreDNS is the default DNS service for Kubernetes that runs in Pods with the label k8s-app=kube-dns. In this lab exercise we'll scale CoreDNS based on the number of schedulable nodes and cores of our cluster. Cluster Proportional Autoscaler will resize the number of CoreDNS replicas.

First lets install CPA using its Helm chart. We'll use the following values.yaml file to configure CPA:

~/environment/eks-workshop/modules/autoscaling/workloads/cpa/values.yaml
options:
  target: deployment/coredns
  namespace: kube-system

config:
  linear:
    nodesPerReplica: 2
    min: 2
    max: 6
    preventSinglePointFailure: true
    includeUnschedulableNodes: true

We summarized how linear scaling works with CPA in the previous section. We can see that we're configuring it to:

  • Target the deployment coredns
  • Add a replica for every 2 worker nodes in the cluster
  • Run at least 2 replicas and to most 6
caution

The configuration above should not be considered best practice for automatically scaling CoreDNS, it is an example that is easy to demonstrate for the purposes of the workshop.

Install the chart:

~$helm repo add cluster-proportional-autoscaler https://kubernetes-sigs.github.io/cluster-proportional-autoscaler
~$helm upgrade --install cluster-proportional-autoscaler cluster-proportional-autoscaler/cluster-proportional-autoscaler \
  --namespace kube-system \
  --version "${CPA_CHART_VERSION}" \
  --set "image.tag=v${CPA_VERSION}" \
  --values ~/environment/eks-workshop/modules/autoscaling/workloads/cpa/values.yaml \
  --wait
NAME: cluster-proportional-autoscaler
LAST DEPLOYED: [...]
NAMESPACE: kube-system
STATUS: deployed
REVISION: 1
TEST SUITE: None


This will create a Deployment in the kube-system namespace which we can inspect:


~$kubectl get deployment cluster-proportional-autoscaler -n kube-system
NAME                              READY   UP-TO-DATE   AVAILABLE   AGE
cluster-proportional-autoscaler   1/1     1            1           92s