Scale Kubernetes Clusters

Cluster Scaling Options

Upon configuring a new cluster, two scaling options are available: manual scaling or autoscaling. Scaling is configured at the node pool level, allowing for more granular control over your cluster’s scaling behaviour.

Once a Kubernetes cluster has been configured, Leaseweb allows you to:

• Adjust the number of nodes within each node pool
• Switch between manual and autoscaling modes for individual node pools as needed

This setup gives you the flexibility to manage workloads per node pool. For example, you can use larger nodes in one pool for resource-critical applications, and smaller nodes in another for lightweight services.
This allows you to optimise performance and cost based on the specific requirements of each workload within the same cluster.

Manual scaling

Enabling Manual Scaling for Node Pools in a Kubernetes Cluster from the Customer Portal

Manual scaling enables you to adjust the number of nodes within each node pool manually, tailored to your specific needs. You have full control over scaling decisions, ensuring each node pool’s size matches the workload requirements at any given time.

The following steps will guide you through scaling your cluster by adjusting node pools as needed:

1. On the Kubernetes Section of the Customer Portal, a list of clusters is visible.
   
2. To configure autoscaling for a node pool, click the Configure action or the cluster’s ID to open the Configure Cluster page.
3. Then, navigate to the Node Pools tab and click edit on the desired node pool.
4. In the edit view, navigate to the Scaling Type option and select Manual Scaling.
5. The desired number of nodes can be chosen in the Amount of Nodes section.
   • Please note that if the selected number of nodes is lower than the currently ready worker nodes, this may lead to instability in your cluster.
     Once satisfied with the new number, click on the update action and then confirm the action with the Confirm button.
6. If the cluster is scaled to fewer nodes, a warning might be emitted.
7. Upon re-confirmation, automation will start scaling the cluster and the node pool accordingly.
   • Cluster admins can monitor the scaling progress within a cluster using the “kubectl get nodes” or refresh the cluster page to visualise the new number of nodes.
   • The scaling process (both scaling up and scaling down) is also reflected in the cluster’s information.
8. You can view the selected scaling type and the desired number of worker nodes within each node pool on the corresponding Node Pools tab.

Back to the top ↑

Autoscaling

Cluster autoscaling automatically adjusts the number of nodes in your cluster based on resource demand. With the introduction of node pools, autoscaling is now configured and managed at the node pool level.
When workloads increase, the system adds nodes to accommodate the load. When demand decreases, it removes unnecessary nodes, ensuring optimal resource usage and cost efficiency.

For more information about cluster autoscaling, refer to the official Kubernetes documentation here.

Enabling Autoscaling for Node Pools in a Kubernetes Cluster from the Customer Portal

You can enable autoscaling by setting minimum and maximum sizes for each node pool.

The following steps will guide you through configuring autoscaling for individual node pools within your cluster as needed:

1. On the Kubernetes Section of the Customer Portal, a list of clusters is visible.
2. To configure autoscaling for a node pool, click the Configure action or the cluster’s ID to open the Configure Cluster page.
3. Then, navigate to the Node Pools tab and click edit on the desired node pool.
4. In the edit view, navigate to the Scaling Type option and select Autoscaling.
   • Minimum amount of Nodes:
     • Defines the smallest size the node pool can scale down to. It must be at least 2 and cannot exceed the Maximum amount of Nodes.
   • Maximum amount of Nodes:
     • Specifies the maximum size the node pool can scale up to. Please note that the maximum amount of workers must be equal to or greater than the currently ready worker nodes. A relevant message will be displayed to inform you of this potential issue.
5. Once satisfied with the new number, click the Update button, then click Confirm to apply the changes.
   • If the cluster is scaled to fewer nodes, a warning might be emitted.
     
6. Upon re-confirmation, automation will start scaling the cluster accordingly.
   • Cluster admins can monitor the scaling progress within a cluster using the “kubectl get nodes” or refresh the cluster page to visualise the new number of nodes.
   • The scaling process (both scaling up and scaling down) is also reflected in the cluster’s information.
7. You can view the selected scaling type and the desired number of worker nodes within each node pool on the corresponding Node Pools tab.

Back to the top ↑

View the Cluster Autoscaling status

To monitor the status of the Autoscaling, you can check the relevant details from the configmap.

Run the following command to retrieve the status of the Autoscaling in the kube-system namespace:

kubectl get configmap cluster-autoscaler-status -o yaml -n kube-system

This command will return an output similar to the example below:

apiVersion: v1
data:
  status: |
    time: 2025-01-09 12:02:54.818691507 +0000 UTC
    autoscalerStatus: Running
    clusterWide:
      health:
        status: Healthy
        nodeCounts:
          registered:
            total: 3
            ready: 3
            notStarted: 0
          longUnregistered: 0
          unregistered: 0
        lastProbeTime: "2025-01-09T12:02:54.818691507Z"
        lastTransitionTime: "2025-01-07T09:29:00.390169767Z"
      scaleUp:
        status: NoActivity
        lastProbeTime: "2025-01-09T12:02:54.818691507Z"
        lastTransitionTime: "2025-01-07T09:29:00.390169767Z"
      scaleDown:
        status: NoCandidates
        lastProbeTime: "2025-01-09T12:02:54.818691507Z"
        lastTransitionTime: "2025-01-07T09:29:00.390169767Z"
. . .

Back to the top ↑

Switching Between Scaling Modes

You can switch between manual and autoscaling modes for your cluster. To do so:

1. To configure autoscaling for a node pool, click the Configure action or the cluster’s ID to open the Configure Cluster page.
   
2. Navigate to the Node Pools tab and click edit on the desired node pool.
3. In the edit view, navigate to the Scaling Type option.
   Then, you change between the manual and automatic scaling options.
   • If the cluster is scaled to fewer nodes, a warning might be emitted.
4. Upon re-confirmation, automation will start scaling the cluster accordingly.
   • Cluster admins can monitor the scaling progress within a cluster using the “kubectl get nodes” or refresh the cluster page to visualise the new number of nodes.
   • The scaling process (both scaling up and scaling down) is also reflected in the cluster’s information.
5. You can view the selected scaling type and the desired number of worker nodes within each node pool on the corresponding Node Pools tab.

Back to the top ↑

Specificity to consider during the scale-down/scale-up of a Kubernetes cluster

• During the BETA release, we currently allow for a minimum of 2 nodes per cluster and a maximum of 10 nodes per cluster.
  • This limit can possibly be increased by opening a support ticket with us.
• We currently do not allow updating the Node offering once a cluster is provisioned.
  • However, you can create a new node pool with your desired node offering, delete the original one, and workloads will be migrated to the new pool.
  • If you have any questions or need additional clarification, please contact us.

Further documentation on Scaling & Related API objects

Name	Short description	Relevant documentation
Disruptions PodDisruptionBudget	Article explaining the PodDisruptionBudget and its use.	*https://kubernetes.io/docs/concepts/workloads/pods/disruptions/ *https://kubernetes.io/docs/tasks/run-application/configure-pdb/
Node affinity	Allows to assign a pod to certain nodes.	*https://kubernetes.io/docs/tasks/configure-pod-container/assign-pods-nodes-using-node-affinity/
NodeSelector	Allow to select on which node a pod can be scheduled	*https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/
Topology	Allow to select on which node a pod can be scheduled	Allows assigning a pod to certain nodes.

Back to the top ↑