Cloud-Native Microservices With Kubernetes - 2nd Edition

Cluster Autoscaler

The Cluster Autoscaler (CA) is a tool that automatically adjusts the size of a Kubernetes cluster in response to the following conditions:

• Pods fail to run in the cluster due to insufficient resources, and
• Nodes in the cluster remain underutilized for an extended period of time, and their Pods can be placed on other existing nodes.

The Cluster Autoscaler is a separate component and has no relation to the HPA, but it could be considered a horizontal autoscaler for the cluster itself. Also, it is not part of the Kubernetes control plane; it is deployed as a Deployment in the kube-system namespace.

Most cloud providers have their own implementation of the Cluster Autoscaler. In most cases, you should use the one provided by your cloud provider since it is optimized for the cloud provider's infrastructure, it is easier to install (usually from the CLI or the web interface), and it is easier to configure.

For example, if we're using DigitalOcean, we can use the web interface to install and configure the Cluster Autoscaler.

Or use the CLI:

doctl kubernetes cluster node-pool update mycluster mypool \
  --auto-scale \
  --min-nodes 1 \
  --max-nodes 10

where:

• mycluster is the name of the cluster
• mypool is the name of the node pool
• auto-scale enables the Cluster Autoscaler
• min-nodes is the minimum number of nodes that the Cluster Autoscaler will scale down to
• max-nodes is the maximum number of nodes that the Cluster Autoscaler will scale up to

You can disable it using:

doctl kubernetes cluster node-pool update mycluster mypool \
  --auto-scale=false

These are some official links to the Cluster Autoscaler documentation and implementations for different cloud providers:

• GCE
• GKE
• AWS
• Azure
• AliCloud
• BaiduCloud
• BizflyCloud