DevOps Engineer interview questions
The 7 questions that actually come up, each with what a strong answer covers. These are not scripts to memorise — interviewers can tell — they are the shape of the answer that gets a yes.
7 questionsDevOps Engineer career guide →
A pod is in CrashLoopBackOff. Diagnose it.
What a strong answer coverskubectl describe + logs (previous), then exit code semantics, probes, resources/OOM, config/secrets. A crisp decision tree wins here.
Design a zero-downtime deployment for a stateful service.
What a strong answer coversRolling with readiness gates, connection draining, backwards-compatible migrations (expand/contract), and rollback that is actually tested.
What makes a GOOD alert?
What a strong answer coversSymptom-based, actionable, tied to user impact, with a runbook. Bonus: alert on SLO burn rate, not on every CPU spike.
Blue-green vs canary vs rolling — when each?
What a strong answer coversCost, rollback speed, and blast radius trade-offs. The trap is reciting definitions without a deciding criterion.
Your pipeline takes 40 minutes. Halve it.
What a strong answer coversMeasure stages first, then parallelise, cache dependencies/layers, split test suites, and question what runs on every commit vs on merge.
How do you manage secrets across environments?
What a strong answer coversA real secret store (Vault, cloud-native), injection at runtime, rotation story, and never in git — plus how you audit access.
Tell me about an incident you handled badly, and what changed after.
What a strong answer coversThey want blameless honesty and a systemic fix (alert, runbook, guardrail), not a hero narrative.