All posts
Kubernetes

The EKS Upgrade Playbook: One Minor Version at a Time, Without an Outage

By

November 2, 2026

6 min read

Kubernetes ships a minor version about every four months and EKS supports each for 14 months, then 12 more at six times the price. A repeatable upgrade: insights first, control plane, add-ons, nodes, then Fargate pods, with a seven-day rollback window for in-place upgrades and the checks between each step.

EKS
EKS Upgrades
Kubernetes
Operations
Karpenter

The cheapest EKS upgrade is the one you do every four months, one minor version at a time. The expensive one is three versions at once in the week before an automatic upgrade, under extended support pricing.

Want a second opinion on this for your system?

Book a free 30-minute call. You'll leave knowing what we'd do first and roughly what it costs, whether or not you hire us.

The order

  1. Read the version's changes in the EKS user guide, and check upgrade insights for deprecated APIs your manifests and Helm charts still use.

  2. Upgrade staging first, and let it run a few days with real traffic patterns.

  3. Upgrade the control plane (one minor version per step). In-place upgrades can be rolled back to the previous minor version within seven days.

  4. Upgrade the managed add-ons: VPC CNI, CoreDNS, kube-proxy, EBS CSI, Pod Identity Agent.

  5. Upgrade nodes: managed node groups by version update, Karpenter by drift and new AMIs, Auto Mode on its own.

  6. Restart Fargate pods (rolling restart of their Deployments) so they come back with the new kubelet.

  7. Upgrade your own controllers: ingress, Argo CD, cert-manager, autoscalers, operators. Their compatibility matrices are usually what blocks a version.

Checks between steps

  • PodDisruptionBudgets on every service that matters, so node replacement drains one replica at a time.

  • Readiness probes that mean "ready", so traffic waits for new pods.

  • Dashboards for error rate and latency open during the change; a written rollback trigger.

  • Nodes never more than one minor version behind the control plane for longer than the upgrade itself.

Making it boring

Practice

Why

A calendar entry per cluster, two months before end of standard support

Extended support starts billing at 00:00 UTC on the date

Clusters defined in code

A staging cluster that matches production is one command

State outside the cluster

Blue-green cluster upgrades become possible when a cluster is disposable

Helm charts and add-ons pinned and updated monthly

Small, frequent bumps instead of a big one at upgrade time

hoolco Monthly Review

Once a month: what changed in tech that month and what it means for you, from cloud and AI to security and the tools you build with. Every item links to its source. No sales pitch, unsubscribe in one click.

As of November 2, 2026. Provided as is, for general information, and accurate to the best of our knowledge on the date shown. Not legal, compliance, tax, financial or other professional advice. Prices, limits and vendor services change; check current vendor documentation before relying on anything here. Terms of use.