3.9.1

Important

Due to REST API version changes, you cannot upgrade to MKE 3.9 directly from MKE 3.8.10 or earlier. Upgrade in the following order:

1. Upgrade to MKE 3.8.11 or later. If your cluster uses MetalLB, upgrade to MKE 3.8.12 or later.

2. Upgrade to MCR 25.0.12 or later.

3. Upgrade to MKE 3.9.

If you are upgrading from MKE 3.7.x, the upgrade to MKE 3.8.x can restart your containers due to the container hash calculation change introduced in Kubernetes 1.31. To prevent containers from restarting simultaneously on all nodes, which the automated in-place cluster upgrade method can cause, use the phased in-place cluster upgrade method or the blue-green deployment upgrade method.

Release date	Name	Highlights
2026-APR-02	MKE 3.9.1	Hotfix release that resolves : Interactive Container Exec failure

Important

If you previously disabled Ingress Controller to mitigate vulnerabilities, Mirantis recommends that following upgrade to MKE 3.9.0 you re-enable the component.

In addition, Mirantis recommends that you migrate from Ingress Controller to Envoy Gateway Controller.

• Enhancements
• Addressed issues
  • [FIELD-8174] Interactive Container Exec failure
• Known issues
  • [MKE-10152] Upgrading large Windows clusters can initiate a rollback
  • [MKE-9699] Ingress Controller with external load balancer can enter crashloop
  • [MKE-8914] Windows Server Core with Containers images incompatible with GCP
  • [MKE-8814] Mismatched MTU values cause Swarm overlay network issues on GCP
  • [MKE-12030] MKE telemetry setting overrides key/value in daemon.json
  • [FIELD-8002] Flag conflict causes worker nodes not to upgrade
  • [FIELD-8196] Missing CustomResourceDefinition (CRD) for MetalLB
  • [MKE-13744] Deprecated –cloud-provider flag causes kube-api-server failure
  • [MKE-13701] Incorrect 0 B image size displays in MKE web UI
  • [FIELD-8241] calico-kube-controllers Pod restarts with leaked IPs
• Major component versions
• Security information