Change a node group

Scale a node group, change its server type, labels, taints or zones, and restart a group that stopped after failed nodes.

Last updated 18 November 2026View as Markdown

Replace my-cluster with your cluster's name or id, and workers with the node group's name.

Prerequisites

  • Permission for eks:UpdateNodegroupConfig on the node group.
  • In the console, also eks:DescribeCluster and eks:ListNodegroups. The console enables Edit only with eks:UpdateNodegroupConfig on every cluster (eks:*).

Change the node group

  1. Open Kubernetes, open the cluster, then choose Node groups.
  2. On the node group's row, choose Edit.
  3. Change the Server type, Minimum, Desired and Maximum, Autoscaling, the labels and the taints as needed.
  4. Choose Save.

What each change does

  • Sizes are checked together with the ones you do not send: minSize ≤ desiredSize ≤ maxSize. Only a larger maxSize is checked against the quotas; shrinking is always allowed.
  • Scaling down removes nodes that are not Ready first, then nodes on an older minor, then the newest. A Ready node is drained first: PodDisruptionBudgets are respected for up to five minutes per node, then the node goes anyway. A node that is not Ready is removed without a drain.
  • Server type applies to nodes created from then on; nodes already running keep theirs until they are replaced.
  • Labels and taints replace the lists as a whole ({} and [] remove them all), and apply to nodes created from then on.
  • Zones apply to nodes created from then on; nothing moves.
  • A group in status failed (stopped after three failed nodes) goes updating on any change and tries again.

Each change is recorded in Activity.

Refusals

  • 400 "Sizes must satisfy minSize ≤ desiredSize ≤ maxSize (would be …)", "Nothing to change", or a type or zone that cannot be used.
  • 404 when the node group does not exist.
  • 409 "The node group is being deleted", or a larger maxSize passes a quota ("Quota exceeded: …").
Did this page help you?