Opscraft
Senior DevOps interviews · Lesson 8 of 10 · 3 min read

Hands-on and live rounds

What to practise

You may be asked to write or fix, live, in 30–45 minutes:

  • A multi-stage Dockerfile for a Java (Maven/Gradle) or Node app — non-root user, small base, good layer caching.
  • Kubernetes manifests: Deployment with probes, resources, securityContext; Service; Ingress; HPA; ConfigMap/Secret usage — possibly with Kustomize overlays.
  • A Jenkinsfile or GitHub Actions workflow for build → test → push → deploy.
  • A bash script: parse logs (top IPs, status codes), check disk usage and alert, retry logic with set -euo pipefail.
  • A small Python script: call an AWS API with boto3 (list untagged resources, old snapshots).
  • A Terraform module: S3 bucket with encryption/versioning/public access block, or a VPC with subnets.
  • Debugging: a broken pod (wrong image, failing probe, missing secret), a Service with no endpoints, a failing pipeline.
Enlarge diagramPipelineJenkinsfile or Actions workflowBashparse logs, check disks, retriesPythonboto3: untagged resources, oldsnapshotsTerraformS3 bucket or VPC moduleDebuggingbroken pod, Service, pipelineTake-homeREADME, how to run, decisions
Figure 1. What live and take-home rounds usually ask for. Practise one of each.
Swipe sideways to see all of it, or tap to open it full screen.

Tips for live rounds

  • Think aloud — explain what you're doing and why.
  • Start simple and working, then improve (security, edge cases).
  • Use docs if allowed; knowing where to look is a skill.
  • Check your work (kubectl apply --dry-run=server, docker build, terraform validate).
  • If stuck, say what you'd try next — partial progress with good reasoning scores well.
  • For take-homes: a clear README, how to run it, design decisions and what you'd do with more time.