Kubernetes interviews blend resource model questions with troubleshooting scenarios: a pod crash looping, high latency after deploy, or autoscaling that never triggers.
Concepts to know cold
- Pod, Deployment, ReplicaSet — desired state and rollouts.
- Service types — ClusterIP, NodePort, LoadBalancer; selectors and endpoints.
- Ingress and TLS termination — routing external traffic to services.
- ConfigMaps and Secrets — configuration vs sensitive material.
- Probes — liveness vs readiness; why bad probes cause outages.
- HPA and resource requests/limits — CPU/memory scheduling and scaling signals.
Troubleshooting narrative
Practice a short debug flow: kubectl get/describe/logs, events, recent deploy diff, dependency health, metrics dashboard. Interviewers want structured thinking, not memorized YAML.
Connect K8s to your resume
Before a live interview, ensure your resume mentions clusters you ran on (GKE, EKS, on-prem), deployment frequency, and one incident — OOM kills, image pull failures, or a bad rollout you rolled back. A copilot surfaces those facts when you get a vague “Tell me about your K8s experience.”