Kubernetes Dns Troubleshooter

Kubernetes DNS Troubleshooter

This skill helps you troubleshoot DNS issues in Kubernetes. DNS is a centerpiece of communication in Kubernetes, it allows pods to discover other applications and communicate with the outside world.

DNS Request Flow

Forward Path (Request):

1. Application Pods → kube-dns Service (DNS request for service/external name)
2. kube-dns Service → CoreDNS Pod (forwarding by host iptables)
3. CoreDNS Pod → Upstream DNS (if external name)
4. Upstream DNS resolves and responds

Return Path (Response):

• Upstream DNS → CoreDNS Pod → kube-dns Service → Application Pods

gadget_trace_dns Tool

Traces DNS requests and responses in real-time using eBPF. It captures:

• Kubernetes context: node, namespace, pod name, container name
• DNS data: query name, query type (A, AAAA, etc.), response code, resolved addresses
• Performance: latency_ns (response time in nanoseconds)
• Network details: source/destination endpoints, nameserver, packet type (HOST/OUTGOING)

Key filtering options:

• operator.KubeManager.namespace - filter by Kubernetes namespace
• operator.KubeManager.podname - filter by pod name
• operator.filter.filter - filter output fields (e.g., latency_ns>1000000 for queries >1ms)

Run modes:

• Foreground (default): returns results after specified duration
• Background: continuous tracing (duration=0)

Troubleshooting Process

1. Check CoreDNS pods are healthy: kubectl get pods -n kube-system -l k8s-app=kube-dns
2. Run gadget_trace_dns filtered by namespace/pod to capture DNS traffic
3. Analyze latency_ns, rcode (response codes), and addresses fields
4. Look for high latency, NXDOMAIN errors (rcode==NameError)
5. Verify DNS requests have corresponding DNS responses
6. If there are any observations, please rerun the tool with additional filtering options.

Source: mayasingh17/ig-mcp-server-scale — distributed by TomeVault.