Cluster Autoscaler is used to scale up or down cluster nodes according to load ( pending pods) or free nodes.
Earlier we had to manually configure Cluster Autoscaling on AKS cluster. Recently Azure AKS released this as a feature. So in this tutorial I would be explaining both the steps.
Manual:
- Generate the secret with the cluster details by running the following script:
https://gist.github.com/anoopl/d0bfb7c78a7a0027934689a7f675796d
2. Run the script and copy the output to a yaml file cluster-autoscaler.yaml:
3. Create the secret :
kubectl apply -f cluster-autoscaler-secret.yaml
4. Use the following yaml to create other resources for the CA:
You can set the minimum and maximum number of nodes in the yaml: — — nodes=1:10:nodepool1
Create the resources with :
kubectl apply -f cluster-autoscaler-all.yaml
You can check the logs to see everything working as expected by checking CA Pod’s logs.
kubectl -n kube-system logs -f <cluster-autoscaler-pod>
Enable Cluster Autoscale feature on AKS
- Update the Az Cli to the latest version:
sudo apt-get update && sudo apt-get install --only-upgrade -y azure-cli
2. Enable aks preview mode
az extension add --name aks-preview
3. Create new cluster with Autoscaling enabled:
You can test the autoscaling by running a deployment of nginx with many replicaset:
Example:
https://gist.github.com/anoopl/891965dd6c9110cdc7950f2d1529041a