DNS Troubleshooting and Custom CoreDNS Configuration

A service call fails with "no such host." The Service exists, the name looks right, kubectl get svc confirms it. You can resolve it from your laptop. But from inside the pod, nothing. No error in the application logs beyond the DNS failure itself. No obvious signal pointing at the root cause.

DNS failures in Kubernetes are uniquely frustrating because they masquerade as application failures. Every service-to-service call flows through CoreDNS, and when that layer breaks, the symptoms appear everywhere except in the DNS infrastructure itself. The CKA exam tests DNS troubleshooting directly under the Services & Networking domain (20%), and the exam tasks map closely to what you'd do in a real outage: spin up a debug pod, check CoreDNS logs, edit the Corefile, fix RBAC.