talos.default service missing

[camaeel]

I have 2 clusters:

• multiple nodes, created ~2-3y ago and frequently upgraded. It has talos.default service present inside the cluster
• 1 node, bare metal (Raspberry PI) clsuter, created couple of days ago. It doesn't have talos.default service.
  Both have this patch applied:

machine:
  features:
    kubernetesTalosAPIAccess:
      enabled: true
      allowedRoles:
        - os:etcd:backup
      allowedKubernetesNamespaces:
        - etcd-backup

[smira]

I don't think we can guess, there is not enough information.

Make sure you're running latest Talos version, and submit a support bundle (talosctl support) if the issue is still there.

[smira]

2026-04-20T21:27:26.876Z �[31mERROR�[0m kubernetes endpoint watch error {"component": "controller-runtime", "controller": "k8s.EndpointController", "error": "failed to list *v1.EndpointSlice: Get \"https://api.talos-mgmnt.kamilk.eu:6443/apis/discovery.k8s.io/v1/namespaces/default/endpointslices?fieldSelector=metadata.name%3Dkubernetes&limit=500&resourceVersion=0\": dial tcp 192.168.189.13:6443: connect: no route to host"}

Your controlplane endpoint doesn't work, which eventually leads to this cascading failure to establish endpoints.

[camaeel]

@smira thank you, probably this is the root cause of the issue (I left somwhere this old IP or it was cached in DNS).

[camaeel]

After some time the service appeared (probably after misconfigured DNS entry TTL expired).