Skip to main content

​
Overview

Scaling a Kubernetes cluster adjusts the worker node count in the default node group. Scale up to handle increased workload demand; scale down to reduce infrastructure cost during low-utilization periods. Node scaling is non-disruptive for scale-up operations. Scale-down operations drain and remove nodes — ensure workloads are properly distributed before reducing node count.
Prerequisites
  • A running cluster in CREATE_COMPLETE or UPDATE_COMPLETE status
  • Sufficient compute quota for the new node count (scale-up only)

​
Scale Up (Add Nodes)

Navigate to the cluster

Log in to the Xloud Dashboard (https://connect.<your-domain>) and navigate to Project → Containers → Clusters.

Resize the cluster

Click Actions → Resize Cluster next to your cluster. Enter the new total node count in the Node Count field.

Confirm

Click Resize. The cluster enters UPDATE_IN_PROGRESS status while new nodes are provisioned and join the cluster.
Cluster returns to UPDATE_COMPLETE once nodes are ready.

​
Scale Down (Remove Nodes)

Scaling down removes nodes from the cluster. Pods running on removed nodes are evicted before the node is deleted. Ensure Pod Disruption Budgets (PDBs) and workload replicas are configured to tolerate node removal without service interruption.

Prepare workloads

Before scaling down, verify that no stateful or single-replica workloads are running on the nodes that will be removed. Use the Kubernetes Dashboard or kubectl to inspect current pod placement.

Resize the cluster

Navigate to Project → Containers → Clusters → Actions → Resize Cluster. Enter the reduced node count and click Resize.

Verify

Wait for the cluster to reach UPDATE_COMPLETE. Confirm remaining nodes are healthy and workloads have been rescheduled.
Cluster is UPDATE_COMPLETE and all pods are running on remaining nodes.

​
Best Practices

Maintain Minimum Replicas

Configure PodDisruptionBudget resources for stateful workloads to ensure a minimum number of replicas remain available during node removal.

Use Node Groups for Fine-Grained Scaling

Use separate node groups for different workload types to scale them independently without affecting other workloads on the cluster.

Scale During Low-Traffic Windows

Perform scale-down operations during low-traffic periods when evicted pods have minimal user impact.

Check Compute Quota Before Scale-Up

Verify sufficient compute quota before adding nodes to avoid partial scale-up failures.

​
Next Steps

Node Groups

Create and manage separate node pools for specialized workloads.

Access Cluster

Verify cluster access after scaling via kubectl.

Cluster Upgrades

Upgrade your cluster to a newer Kubernetes version.

Troubleshooting

Resolve scaling failures and node health issues.