When your cluster's resource demands shift, you can resize worker nodes in Container Service for Kubernetes (ACK) by changing their instance types — scaling up for heavier workloads or scaling down to reduce costs.
Back up node data before changing instance types. The node stays in the Unschedulable state throughout the process. During draining, ACK evicts pods based on configured PodDisruptionBudgets (PDBs). To maintain availability, use a multi-replica deployment strategy so workloads are distributed across multiple nodes. You can also configure PDBs for key services to control the number of pods that are interrupted at the same time. If the node pool has node scaling enabled, temporarily set Scaling Mode to Manual before resizing, then re-enable it afterward.
Before choosing an instance type, see Suggestions on choosing ECS specifications for ACK clusters.
Prerequisites
Before you begin, ensure that you have:
Access to the ACK console and the ECS console
Change the instance type of a single node
These steps apply to subscription worker nodes. For billing details, see Overview of instance configuration changes.
Log on to the ACK console. In the left navigation pane, click Clusters.
On the Clusters page, click the cluster name. In the left navigation pane, choose Nodes > Nodes.
Find the node and choose More > Drain in the Actions column. In the Drain message, click Confirm.
After the node is drained, click the instance ID to open the instance details page. On the Instance Details tab, locate the Configuration Information section and click Change next to the instance type. The Change Instance Type dialog box appears.
Select the target instance type, review the estimated cost, and click Confirm Change and Pay.
In the Stop Instance dialog box, set the Stopped By parameter and click OK. For details, see Stop an instance.
After the configuration change completes, click Start Instance and Return. In the Start Instance dialog box, click OK. Wait for the node to re-join the cluster and enter the Ready state. Then choose Schedulability Settings in the Actions column to set the node back to schedulable.
Change the instance type of multiple nodes in batches
The behavior differs depending on whether the nodes belong to the same node pool:
| Scenario | ECS instances | Custom node configurations |
|---|---|---|
| Nodes from different node pools | Retained (not replaced) | Retained |
| Nodes from the same node pool | Replaced by new instances | Deleted |
Nodes from different node pools
Original ECS instances are not replaced and custom node configurations are retained.
Log on to the ACK console. In the left navigation pane, click Clusters.
On the Clusters page, click the cluster name. In the left navigation pane, choose Nodes > Nodes.
Select the nodes and click Drain at the bottom of the page. In the Drain dialog box, click OK.
After the nodes are drained, log on to the ECS console. In the left navigation pane, choose Instances & Images > Instances.
Select the target ECS instances and choose More > Upgrade/Downgrade > Change Instance Type at the bottom of the page.
In the Destination Instance Type column, select a new instance type for each node and click Next.
Check instance states and click Next:
If all instances are Stopped, click Next.
If any instance is Running, click Stop All, then click Next.
ImportantFor pay-as-you-go instances, set Stop Mode to Standard Mode. Economical Mode reclaims vCPUs and memory, and the instance may fail to restart if resources are insufficient. Stop instances during off-peak hours to minimize service impact.
Review the cost and click Confirm Change.
Start the instances. After they re-join the cluster and enter the Ready state, select the nodes and click Set Node Schedulability at the bottom of the page.
Nodes from the same node pool
This process replaces the original ECS instances with new ones. Custom configurations on the original nodes are deleted.
Log on to the ACK console. In the left navigation pane, click Clusters.
On the Clusters page, click the cluster name. In the left navigation pane, choose Nodes > Node Pools.
Find the node pool and click Edit in the Actions column. Update the instance type in the Instance Type section and click Confirm.
Manually scale out the node pool to add nodes of the new instance type.
After the new nodes enter the Ready state, choose Nodes > Nodes in the left navigation pane.
Find the original nodes and choose More > Drain in the Actions column. In the Drain message, click Confirm.
Confirm that workloads are running correctly on the new nodes, then remove the original nodes.