Kubernetes Debugging Skill

Overview

Systematic toolkit for debugging Kubernetes clusters, workloads, networking, and storage with a deterministic, safety-first workflow.

Trigger Phrases

Use this skill when requests resemble:

• "My pod is in CrashLoopBackOff; help me find the root cause."
• "Service DNS works in one pod but not another."
• "Deployment rollout is stuck."
• "Pods are Pending and not scheduling."
• "Cluster health looks degraded after a change."
• "PVC is pending and pods cannot mount storage."

Prerequisites

Run from the skill directory (devops-skills-plugin/skills/k8s-debug) so relative script paths work as written.

Required

• kubectl installed and configured.
• An active cluster context.
• Read access to namespaces, pods, events, services, and nodes.

Quick preflight:

kubectl config current-context
kubectl auth can-i get pods -A
kubectl auth can-i get events -A
kubectl get ns

Optional but Recommended

• jq for more precise filtering in ./scripts/cluster_health.sh.
• Metrics API (metrics-server) for kubectl top.
• In-container debug tools (nslookup, getent, curl, wget, ip) for deep network tests.

Fallback behavior:

• If optional tools are missing, scripts continue and print warnings with reduced output.
• If kubectl top is unavailable, continue with kubectl describe and events.

When to Use This Skill

Use this skill for:

• Pod failures (CrashLoopBackOff, ImagePullBackOff, Pending, OOMKilled)
• Service connectivity or DNS resolution issues
• Network policy or ingress problems
• Volume and storage mount failures
• Deployment rollout issues
• Cluster health or performance degradation
• Resource exhaustion (CPU/memory)
• Configuration problems (ConfigMaps, Secrets, RBAC)

Safety Rules for Disruptive Commands

Default mode is read-only diagnosis first. Only execute disruptive commands after confirming blast radius and rollback.

Commands requiring explicit confirmation:

• kubectl delete pod ... --force --grace-period=0
• kubectl drain ...
• kubectl rollout restart ...
• kubectl rollout undo ...
• kubectl debug ... --copy-to=...

Before disruptive actions:

# Snapshot current state for rollback and incident notes
kubectl get deploy,rs,pod,svc -n \x3Cnamespace> -o wide
kubectl get pod \x3Cpod-name> -n \x3Cnamespace> -o yaml > before-\x3Cpod-name>.yaml
kubectl get events -n \x3Cnamespace> --sort-by='.lastTimestamp' > before-events.txt

Reference Navigation Map

Load only the section needed for the observed symptom.

Scripts Overview

Script Exit Codes

./scripts/cluster_health.sh and ./scripts/network_debug.sh share the same contract:

• 0: checks completed with no check failures (warnings allowed unless --strict is set).
• 1: one or more checks failed, or warnings occurred in --strict mode.
• 2: blocked preconditions (for example: missing kubectl, no active context, inaccessible namespace/pod).

Deterministic Debugging Workflow

Follow this systematic approach for any Kubernetes issue:

1. Preflight and Scope

kubectl config current-context
kubectl get ns
kubectl auth can-i get pods -n \x3Cnamespace>

If preflight fails, stop and fix access/context first.

2. Identify the Problem Layer

Categorize the issue:

• Application Layer: Application crashes, errors, bugs
• Pod Layer: Pod not starting, restarting, or pending
• Service Layer: Network connectivity, DNS issues
• Node Layer: Node not ready, resource exhaustion
• Cluster Layer: Control plane issues, API problems
• Storage Layer: Volume mount failures, PVC issues
• Configuration Layer: ConfigMap, Secret, RBAC issues

3. Gather Diagnostics with the Right Script

Use the appropriate diagnostic script based on scope:

Pod-Level Diagnostics

Use ./scripts/pod_diagnostics.py for comprehensive pod analysis:

python3 ./scripts/pod_diagnostics.py \x3Cpod-name> -n \x3Cnamespace>

This script gathers:

• Pod status and description
• Pod events
• Container logs (current and previous)
• Resource usage
• Node information
• YAML configuration

Output can be saved for analysis:

python3 ./scripts/pod_diagnostics.py \x3Cpod-name> -n \x3Cnamespace> -o diagnostics.txt

Cluster-Level Health Check

Use ./scripts/cluster_health.sh for overall cluster diagnostics:

./scripts/cluster_health.sh > cluster-health-$(date +%Y%m%d-%H%M%S).txt

This script checks:

• Cluster info and version
• Node status and resources
• Pods across all namespaces
• Failed/pending pods
• Recent events
• Deployments, services, statefulsets, daemonsets
• PVCs and PVs
• Component health
• Common error states (CrashLoopBackOff, ImagePullBackOff)

Network Diagnostics

Use ./scripts/network_debug.sh for connectivity issues:

./scripts/network_debug.sh \x3Cnamespace> \x3Cpod-name>
# or force warning sensitivity / insecure TLS only when explicitly needed:
./scripts/network_debug.sh --strict \x3Cnamespace> \x3Cpod-name>
./scripts/network_debug.sh --insecure \x3Cnamespace> \x3Cpod-name>

This script analyzes:

• Pod network configuration
• DNS setup and resolution
• Service endpoints
• Network policies
• Connectivity tests
• CoreDNS logs

4. Follow Issue-Specific Reference Workflow

Based on the identified issue, consult ./references/troubleshooting_workflow.md:

• Pod Pending: Resource/scheduling workflow
• CrashLoopBackOff: Application crash workflow
• ImagePullBackOff: Image pull workflow
• Service issues: Network connectivity workflow
• DNS failures: DNS troubleshooting workflow
• Resource exhaustion: Performance investigation workflow
• Storage issues: PVC binding workflow
• Deployment stuck: Rollout workflow

5. Apply Targeted Fixes

Refer to ./references/common_issues.md for symptom-specific fixes.

6. Verify and Close

Run final verification:

kubectl get pods -n \x3Cnamespace> -o wide
kubectl get events -n \x3Cnamespace> --sort-by='.lastTimestamp' | tail -20
kubectl rollout status deployment/\x3Cname> -n \x3Cnamespace>

Issue is done when user-visible behavior is healthy and no new critical warning events appear.

Example Flows

Example 1: CrashLoopBackOff in payments Namespace

python3 ./scripts/pod_diagnostics.py payments-api-7c97f95dfb-q9l7k -n payments -o payments-diagnostics.txt
kubectl logs payments-api-7c97f95dfb-q9l7k -n payments --previous --tail=100
kubectl get deploy payments-api -n payments -o yaml | grep -A 8 livenessProbe

Then open ./references/common_issues.md and apply the CrashLoopBackOff solutions.

Example 2: Service DNS/Connectivity Failure

./scripts/network_debug.sh checkout checkout-api-75f49c9d8f-z6qtm
kubectl get svc checkout-api -n checkout
kubectl get endpoints checkout-api -n checkout
kubectl get networkpolicies -n checkout

Then follow Service Connectivity Workflow in ./references/troubleshooting_workflow.md.

Essential Manual Commands

Pod Debugging

# View pod status
kubectl get pods -n \x3Cnamespace> -o wide

# Detailed pod information
kubectl describe pod \x3Cpod-name> -n \x3Cnamespace>

# View logs
kubectl logs \x3Cpod-name> -n \x3Cnamespace>
kubectl logs \x3Cpod-name> -n \x3Cnamespace> --previous  # Previous container
kubectl logs \x3Cpod-name> -n \x3Cnamespace> -c \x3Ccontainer>  # Specific container

# Execute commands in pod
kubectl exec \x3Cpod-name> -n \x3Cnamespace> -it -- /bin/sh

# Get pod YAML
kubectl get pod \x3Cpod-name> -n \x3Cnamespace> -o yaml

Service and Network Debugging

# Check services
kubectl get svc -n \x3Cnamespace>
kubectl describe svc \x3Cservice-name> -n \x3Cnamespace>

# Check endpoints
kubectl get endpoints -n \x3Cnamespace>

# Test DNS
kubectl exec \x3Cpod-name> -n \x3Cnamespace> -- nslookup kubernetes.default

# View events
kubectl get events -n \x3Cnamespace> --sort-by='.lastTimestamp'

Resource Monitoring

# Node resources
kubectl top nodes
kubectl describe nodes

# Pod resources
kubectl top pods -n \x3Cnamespace>
kubectl top pod \x3Cpod-name> -n \x3Cnamespace> --containers

Emergency Operations

# Restart deployment
kubectl rollout restart deployment/\x3Cname> -n \x3Cnamespace>

# Rollback deployment
kubectl rollout undo deployment/\x3Cname> -n \x3Cnamespace>

# Force delete stuck pod
kubectl delete pod \x3Cpod-name> -n \x3Cnamespace> --force --grace-period=0

# Drain node (maintenance)
kubectl drain \x3Cnode-name> --ignore-daemonsets --delete-emptydir-data

# Cordon node (prevent scheduling)
kubectl cordon \x3Cnode-name>

Completion Criteria

Troubleshooting session is complete when all are true:

• Cluster context and namespace are confirmed.
• Relevant diagnostic script output is captured.
• Root cause is identified and tied to evidence (events/logs/config/state).
• Any disruptive action was preceded by snapshot and rollback plan.
• Fix verification commands show healthy state.
• Reference path used (./references/troubleshooting_workflow.md or ./references/common_issues.md) is documented in notes.

Related Tools

Useful additional tools for Kubernetes debugging:

• kubectl-debug: Advanced debugging plugin
• stern: Multi-pod log tailing
• kubectx/kubens: Context and namespace switching
• k9s: Terminal UI for Kubernetes
• lens: Desktop IDE for Kubernetes
• Prometheus/Grafana: Monitoring and alerting
• Jaeger/Zipkin: Distributed tracing