Kubernetes Retires Its Own Ingress NGINX Controller as Part of the 1.35 Release Cycle

Alongside the Kubernetes 1.35 "Timbernetes" release, the project confirmed that the official Ingress NGINX controller, long the default, most widely deployed way to route external traffic into a cluster, is moving to best-effort maintenance only, with full retirement scheduled for March 2026. For any platform or infrastructure team that adopted Ingress NGINX years ago simply because it was the obvious, community-blessed default, this forces a migration decision that a lot of teams have been able to defer indefinitely until now. The retirement is the tail end of a longer transition the Kubernetes project has been steering toward for a couple of years: the newer Gateway API, which graduated to general availability and has continued maturing through 2025 and 2026, was explicitly designed to replace the aging Ingress resource with a more expressive, extensible routing model, and a growing list of implementations, Envoy Gateway, Istio, Cilium, and others, now support it as a first-class citizen rather than a bolt-on. Kubernetes 1.35 itself also graduated in-place Pod resource updates to general availability, letting CPU and memory limits change without restarting a Pod, and made the image volume type stable, letting a Pod pull an OCI image from a registry and mount its contents directly as a read-only volume, both meaningful for teams running dense, resource-sensitive workloads. But the Ingress NGINX retirement is the change that actually forces action: teams still relying on it should treat this as a hard deadline to evaluate a Gateway API implementation, since running an officially unmaintained ingress controller in production is a growing security and reliability liability that only gets worse the longer it's deferred.

Source

View on ShipDigest