Autoscaling Kubernetes clusters is essential to optimizing performance and reducing costs. Cluster Autoscaler (CA) helps you dynamically scale nodes to meet changing workloads, reducing resource waste while keeping applications reliable. In this post, we’ll explore how CA works, how it compares with Horizontal Pod Autoscaler (HPA) and Vertical Pod Autoscaler (VPA), and walk through implementing CA in Kubernetes using Rancher.

Understanding the Different Types of Autoscaling in Kubernetes

Kubernetes supports three key types of autoscaling:

  1. Vertical Pod Autoscaler (VPA):
    Adjusts the CPU and memory requests for individual pods to optimize resource usage. It ensures your applications always get the resources they need to run efficiently.

  2. Horizontal Pod Autoscaler (HPA):
    Increases or decreases the number of pod replicas based on metrics like CPU usage or custom metrics. HPA helps maintain application performance during sudden spikes in demand.

  3. Cluster Autoscaler (CA):
    Scales the number of nodes in your Kubernetes cluster to ensure it can accommodate the workloads running inside it. CA adds or removes nodes based on resource demands, optimizing cluster utilization.

While VPA and HPA manage resources at the pod level, CA scales the entire cluster to meet your application’s changing needs.

Prerequisites for Setting Up Cluster Autoscaler

You can use Rancher to easily manage CA for your Kubernetes clusters. For this tutorial, you’ll need:

  • Rancher installed (via K3s, Docker, Rancher Desktop, or a cloud provider)
  • A Linode account for deploying and testing Cluster Autoscaler on Linode Kubernetes Engine (LKE)
  • A personal access token from Linode to authenticate with the LKE API

Deploying a Kubernetes Cluster with Rancher

Once Rancher is installed, follow these steps to deploy a Kubernetes cluster with Cluster Autoscaler enabled:

  1. Log in to Rancher and go to Cluster Management.
  2. Activate the LKE cluster driver by selecting Drivers → Activate.
  3. Navigate to Clusters → Create, choose Linode LKE, and provide the required information such as cluster name and access token.
  4. Select your region and Kubernetes version, then configure node pools:
    • Example: One node pool with 2GB nodes and another with 4GB nodes.
  5. Click Create to provision your cluster.

Rancher uses Cluster API to deploy and manage your cluster, providing seamless automation for resource scaling.

Configuring Cluster Autoscaler on Rancher

Step 1: Create a Namespace for CA

  1. In Rancher, go to Projects/Namespaces → Create Namespace.
  2. Create a new namespace for Cluster Autoscaler.

Step 2: Create a Secret for Node Group Configuration

Use the following YAML to create a secret for your CA configuration:

apiVersion: v1
kind: Secret
metadata:
  name: cluster-autoscaler-cloud-config
  namespace: autoscaler
type: Opaque
stringData:
  cloud-config: |-
    [global]
    linode-token=<PERSONAL_ACCESS_TOKEN>
    lke-cluster-id=88612
    default-min-size-per-linode-type=1
    default-max-size-per-linode-type=5

    [nodegroup "g6-standard-1"]
    min-size=1
    max-size=4

    [nodegroup "g6-standard-2"]
    min-size=1
    max-size=2

This configuration defines the Linode node groups and their scaling limits. Replace <PERSONAL_ACCESS_TOKEN> with your Linode token.

Step 3: Deploy Cluster Autoscaler

In Rancher, use the import YAML feature to deploy the following CA components:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: cluster-autoscaler
  namespace: autoscaler
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: cluster-autoscaler
    spec:
      containers:
      - name: cluster-autoscaler
        image: k8s.gcr.io/autoscaling/cluster-autoscaler:v1.26.1
        command:
          - ./cluster-autoscaler
          - --cloud-provider=linode
          - --cloud-config=/config/cloud-config
        volumeMounts:
          - name: cloud-config
            mountPath: /config
        volumes:
          - name: cloud-config
            secret:
              secretName: cluster-autoscaler-cloud-config

This deployment will create the Cluster Autoscaler and link it to the Linode node groups you defined in the secret.

Testing Cluster Autoscaler

Step 1: Scale Up with a Workload

Deploy the following busybox workload with 600 replicas:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: busybox-workload
spec:
  replicas: 600
  template:
    spec:
      containers:
      - name: busybox
        image: busybox
        command: ['sh', '-c', 'echo Demo Workload ; sleep 600']

Watch as Cluster Autoscaler detects the increased demand and adds nodes to your cluster to accommodate the workload.

Step 2: Scale Down

Delete the workload from the Rancher UI. Cluster Autoscaler will automatically remove unused nodes after a few minutes, optimizing resource usage.

Best Practices for Cluster Autoscaler

  1. Define Node Groups Carefully: Organize node groups by workload type to optimize scaling.
  2. Monitor with Prometheus: Use Prometheus and Grafana to monitor CA’s performance and detect scaling issues.
  3. Test Scaling Regularly: Ensure that CA can respond to workload changes by running periodic tests.
  4. Set Clear Limits: Define min and max limits for each node group to control scaling behavior.

Conclusion

Cluster Autoscaler is an essential tool for managing Kubernetes clusters efficiently. With Rancher, setting up and managing CA becomes simple, providing automatic node scaling that ensures your clusters can handle any workload.

By leveraging CA, Rancher, and Cluster API, you can create scalable, cost-efficient infrastructure that adapts to your organization’s needs. Implementing autoscaling with Rancher ensures that your applications remain performant, while reducing manual intervention and minimizing costs.