eks-diagnostics

name	eks-diagnostics
description	Use this skill to investigate and troubleshoot Amazon EKS problems by analyzing cluster configurations, node health, pod scheduling, networking, IAM/RBAC, storage, scaling, add-ons, and following structured runbooks. Activate when: cluster creation failures, node not ready, pod pending or CrashLoopBackOff, OOMKilled, image pull errors, VPC CNI issues, CoreDNS failures, service connectivity problems, ingress/ALB issues, IRSA failures, aws-auth ConfigMap misconfiguration, RBAC denied, pod identity issues, EBS/EFS CSI driver problems, PVC pending, cluster autoscaler or Karpenter issues, HPA not scaling, add-on failures, version compatibility, cluster upgrades, or the user says something is wrong with EKS without naming specific symptoms.
compatibility	Requires AWS CLI, kubectl, eksctl, CloudWatch (Container Insights), and CloudTrail permissions. Optional: Helm, Karpenter CLI.

EKS Diagnostics

When to use

Any EKS investigation where the console alone is insufficient — cluster lifecycle issues, node health, pod scheduling failures, networking problems, IAM/RBAC misconfigurations, storage driver issues, autoscaling failures, add-on management, or upgrade planning.

Investigation workflow

Step 1 — Collect and triage

# Cluster status, version, endpoint, logging configuration
aws eks describe-cluster --name <cluster> --query 'cluster.{status:status,version:version,endpoint:endpoint,logging:logging,platformVersion:platformVersion}'

# Node health and readiness
kubectl get nodes -o wide

# Pods not in Running state across all namespaces
kubectl get pods --all-namespaces --field-selector status.phase!=Running

# Node conditions and resource pressure
kubectl describe node <node-name>

# Recent cluster events sorted by timestamp
kubectl get events --sort-by='.lastTimestamp' -A | tail -50

Triage returns: