Report #103429
[bug\_fix] DNS resolution fails inside pods for cluster services
Verify CoreDNS is running \(\`kubectl get pods -n kube-system -l k8s-app=kube-dns\`\) and that the cluster DNS Service exists \(\`kubectl get svc -n kube-system\`\). Check the pod's \`/etc/resolv.conf\` and the kubelet \`--cluster-dns\` / \`clusterDNS\` config to ensure it points at the CoreDNS Service IP. If CoreDNS pods are crash-looping, inspect their logs and memory limits. If NetworkPolicy blocks UDP/TCP port 53, allow DNS traffic. For custom domain suffixes, confirm \`ndots\` and search domains in resolv.conf.
Journey Context:
From inside a pod, \`curl http://my-service\` times out, but \`curl http://my-service.my-namespace.svc.cluster.local\` works — or nothing resolves at all. This points to Kubernetes DNS, which is implemented by CoreDNS in most clusters. Start with \`kubectl get pods -n kube-system -l k8s-app=kube-dns\`; if CoreDNS pods are not running or are in CrashLoopBackOff, the whole cluster loses service discovery. Read the CoreDNS logs: common causes are OOMKilled due to a low memory limit under high query load, or a Corefile syntax error from a custom config. If CoreDNS is healthy, exec into a failing pod and run \`cat /etc/resolv.conf\`; the \`nameserver\` should be the cluster DNS IP \(the CoreDNS Service IP\) and the search path should include \`svc.cluster.local\`. Mismatches happen when a cluster is rebuilt and the kubelet still has the old DNS IP, or when a custom CNI/NetworkPolicy blocks port 53. Fix the CoreDNS deployment resources, correct the kubelet clusterDNS config, or update NetworkPolicy to allow egress to the DNS service, and pod DNS resolves again.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-07-11T04:23:14.870820+00:00— report_created — created