Check manifests against the Kubernetes API versions that have been removed, or browse the full list. States which Kubernetes release it covers.
Waiting for a manifest
Paste manifests to find the ones a current cluster will reject outright.
Removed versions appear here
A removed API version is not a warning — the endpoint is gone, and the API server returns an error that reads like a missing resource type.
This is maintained reference data written against Kubernetes 1.32, and it has not asked your cluster anything. A managed provider may keep an API serving past its upstream removal, or retire one early; a cluster with custom resources has versions this list has never heard of. For what your cluster actually serves, run kubectl api-resources, and for what it is about to stop serving, kubectl get --raw /metrics exposes an apiserver_requested_deprecated_apis counter that names the callers.
Kubernetes API Deprecations runs entirely on your device — no cluster is contacted and your manifests are never uploaded.
You’re on the Free plan. Compare plans
Kubernetes does not merely warn about an old API version and keep serving it. It removes the endpoint. An `extensions/v1beta1` Ingress applied to a cluster running 1.22 or later is not deprecated — it is a 404 from the API server, and the error says the resource type does not exist, which reads like a missing CRD rather than a version problem.
That is why a cluster upgrade can break a manifest nobody has touched in three years, and why the failure is so often misdiagnosed. The manifest was correct when it was written and is now addressing an endpoint that is gone.
This is maintained static reference data and it states which Kubernetes release it was written against, because that is the only honest way to publish a list like this. It has not asked your cluster anything.
Your cluster may differ in both directions. A managed provider sometimes keeps an API serving past its upstream removal, or retires one early; a cluster with custom resources has versions this list has never heard of. For what your cluster actually serves, `kubectl api-resources` is the authority, and `kubectl get --raw /metrics` exposes an `apiserver_requested_deprecated_apis` counter that names the callers still using a doomed version.
Most likely it uses an API version the upgrade removed. Kubernetes deletes the endpoint rather than deprecating it, so the API server returns an error that looks like a missing resource type. Paste the manifest here and the removed versions are listed with their replacements.
It states the Kubernetes release it was written against, and it has not contacted your cluster. Managed providers sometimes keep an API alive past upstream removal or retire one early, so treat `kubectl api-resources` as the authority for what yours actually serves.
Usually but not always. Most replacements take the same fields — `batch/v1beta1` to `batch/v1` for a CronJob is just the line. The Ingress move to `networking.k8s.io/v1` changed the backend shape and made `pathType` required, and Pod Security Policy has no drop-in replacement at all. Each entry says which it is.