Migrate EndpointSlice from discovery.k8s.io/v1beta1

EndpointSlice tracks the network endpoints behind a Service. Kubernetes 1.25 removed discovery.k8s.io/v1beta1. Most often, the control plane creates and manages EndpointSlices, but anything that writes its own, such as a service mesh, must move the topology entries to the new nodeName and zone fields.

The error

After the upgrade to 1.25, kubectl apply fails like this:

error: resource mapping not found for name: "<name>" namespace: "<ns>" from "<file>": no matches for kind "EndpointSlice" in version "discovery.k8s.io/v1beta1"
ensure CRDs are installed first

What replaces it

Replacement: discovery.k8s.io/v1
Available since: 1.21
Deprecated in: 1.21
Removed in: 1.25

Check a manifest against 1.25, or read the official documentation.

Migration notes

use per Endpoint nodeName field instead of deprecated topology["kubernetes.io/hostname"] field
use per Endpoint zone field instead of deprecated topology["topology.kubernetes.io/zone"] field
topology is replaced with the deprecatedTopology field which is not writable in v1

Before and after

Before:

apiVersion: discovery.k8s.io/v1beta1
kind: EndpointSlice
metadata:
  name: web-1
  labels:
    kubernetes.io/service-name: web
addressType: IPv4
ports:
  - name: http
    protocol: TCP
    port: 80
endpoints:
  - addresses:
      - "10.1.2.3"
    topology:
      kubernetes.io/hostname: node-1
      topology.kubernetes.io/zone: zone-a

After:

apiVersion: discovery.k8s.io/v1
kind: EndpointSlice
metadata:
  name: web-1
  labels:
    kubernetes.io/service-name: web
addressType: IPv4
ports:
  - name: http
    protocol: TCP
    port: 80
endpoints:
  - addresses:
      - "10.1.2.3"
    nodeName: node-1
    zone: zone-a

Helm upgrade fails after a cluster upgrade

If you installed this with Helm, helm upgrade can fail after the cluster is upgraded, even if the chart is already fixed. Which error appears depends on your Helm version and the error text:

Error: UPGRADE FAILED: unable to build kubernetes objects from current release manifest: resource mapping not found for name: "<name>" namespace: "<ns>" from "": no matches for kind "EndpointSlice" in version "discovery.k8s.io/v1beta1"

Error: UPGRADE FAILED: current release manifest contains removed kubernetes api(s) for this kubernetes version and it is therefore unable to build the kubernetes objects for performing the diff. error from kubernetes: unable to recognize "": no matches for kind "EndpointSlice" in version "discovery.k8s.io/v1beta1"

Helm compares your chart with the manifest stored with the release. That stored manifest still has the old API, and the cluster can no longer read it.

To check, get the stored manifest, and paste the output into the checker:

helm get manifest <release> -n <namespace>

To fix the stored record, use the helm-mapkubeapis plugin, latest tag v0.6.1. Try --dry-run first. Its README lists support for Helm v3, so check that it suits your Helm version.

helm plugin install https://github.com/helm/helm-mapkubeapis --version v0.6.1
helm mapkubeapis <release> --namespace <namespace> --dry-run

Helm's advice is to move releases to supported APIs before you upgrade the cluster, and not to roll back to a release from before that change.