Redistribute coreDNS pods after fresh cluster install

Description

When bootstrapping an HA Talos cluster, both CoreDNS replicas may initially be scheduled on the same control plane node. If that node crashes, DNS resolution is not available until the pod(s) are rescheduled.

The Kubernetes kubeadm HA documentation recommends restarting CoreDNS after the additional nodes have joined:

• https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/high-availability/#steps-for-the-rest-of-the-control-plane-nodes

kubectl rollout restart deployment -n kube-system coredns

This allows the scheduler to redistribute the CoreDNS replicas across the available nodes.

It might be useful to add a similar note to the Talos HA/production cluster documentation.

Related Kubernetes documentation issue: kubernetes/website#52089

References

• https://github.com/kubernetes/website/issues/52089
• https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/high-availability/#steps-for-the-rest-of-the-control-plane-nodes

Start with the Talos HA/production cluster documentation and compare its guidance with the linked Kubernetes kubeadm high-availability section. Add a note explaining when to restart the CoreDNS deployment after additional control plane nodes join, then verify that the documented command and placement are clear.