DevOps and SRE interview questions
41 questions asked in DevOps and SRE interviews, on infrastructure, reliability, on-call and CI/CD. They are built from thousands of real job ads. Under each one: what the interviewer is listening for, and the follow-up they may ask. For the recruiter call, see the HR screen questions.
Behavioral questions
Questions about how you work in this role. More in behavioral questions for any tech role.
-
1
Tell me about a time a change you made caused a problem in production.
All levels
Listening for Owns it quickly, helps fix it, and changes how they deploy afterwards.
Follow-up Ask what safety check they added after that.
-
2
Tell me about a time someone asked you to skip a safety step to deploy faster. What did you do?
Junior, Mid
Listening for Holds the line politely or finds a safe shortcut.
Follow-up Ask how the other person reacted.
-
3
Tell me about a time you had to keep people updated during an incident. How did you do it?
All levels
Listening for Short, regular, honest updates for the right audience.
Follow-up Ask what they said when they did not know the cause yet.
-
4
Tell me about a time you wrote a runbook or docs that other people actually used.
Junior, Mid
Listening for Writes for the reader under stress and keeps it current.
Follow-up Ask how they found out it was useful.
-
5
Tell me about a time developers pushed back on a tool or process your team introduced. How did you get them on board?
Mid, Senior
Listening for Listens to the real complaint and adjusts instead of forcing it.
Follow-up Ask what they changed because of the feedback.
-
6
Tell me about a time on-call was wearing the team out. What did you change?
Mid, Senior
Listening for Notices the human cost and fixes causes, not just the rota.
Follow-up Ask how they knew the change worked.
-
7
Tell me about a big migration you led, like moving to the cloud or to Kubernetes. How did you keep things running?
Senior
Listening for Plans in steps with rollback, brings teams along, tracks risk.
Follow-up Ask what went wrong and how they handled it.
Technical discussion questions
Talking through your experience and decisions, not a coding test. Grouped by topic.
Observability
-
1
Can you explain the difference between logs, metrics and traces, and when you'd look at each?
Junior
Listening for Understands each in their own words with an example.
Follow-up Ask which one they used most recently and for what.
-
2
Our on-call engineers get forty alerts a night, and most are noise. How would you fix that?
Mid, Senior
Listening for Alerts on user impact, deletes or tunes noisy ones, tracks alert quality.
Follow-up Ask how they would convince owners to delete an alert.
-
3
A team is launching a new service next month. What would you ask them to set up for monitoring before launch?
Mid, Senior
Listening for Key metrics, alerts tied to users, dashboards, logs with context, a runbook.
Follow-up Ask which of these they would never skip.
-
4
A product team wants 100 percent uptime for their service. How would you set a realistic target with them?
Mid, Senior
Listening for Explains SLOs and error budgets in plain words, ties them to users and cost.
Follow-up Ask what happens when the budget runs out.
Kubernetes
-
5
A pod in Kubernetes keeps restarting. How would you find out why?
Junior, Mid
Listening for Checks events, logs from the last run, exit codes, probes and limits.
Follow-up Ask what they do if the logs are empty.
-
6
How would you roll out a change to a busy service on Kubernetes without downtime?
Mid, Senior
Listening for Rolling or canary deploys, readiness probes, graceful shutdown, compatible changes.
Follow-up Ask how they handle a database change in the same release.
-
7
How would you plan a Kubernetes version upgrade across several production clusters?
Senior
Listening for Reads breaking changes, tests on a low-risk cluster first, has a rollback plan and talks to app teams.
Follow-up Ask what they would do if one team is not ready.
AWS
-
8
Our cloud bill went up thirty percent last month. How would you find out why?
All levels
Listening for Breaks costs down by service and team, looks for leftovers and traffic changes.
Follow-up Ask how they would stop it happening again.
-
9
Leadership wants to run on both AWS and GCP. What would you tell them about the trade-offs?
Senior
Listening for Honest about cost, complexity and skills versus lock-in and resilience.
Follow-up Ask what they would suggest instead if the real goal is uptime.
CI/CD
-
10
Walk me through a CI/CD pipeline you've worked on. Which step breaks most often?
Junior, Mid
Listening for Knows each stage and why it is there, and real failure points.
Follow-up Ask what they did about the step that breaks.
-
11
A deploy you just did is causing errors. How do you decide between rolling back and fixing forward?
Junior, Mid
Listening for Restores users first, rolls back unless that is risky, then fixes calmly.
Follow-up Ask when a rollback itself is dangerous.
-
12
Our CI builds take forty minutes and developers are unhappy. How would you speed them up?
Mid, Senior
Listening for Measures where time goes, then caching, parallel steps, smaller images.
Follow-up Ask how they would keep it fast over time.
Terraform
-
13
In your own words, why would a team manage cloud resources with Terraform instead of clicking in the console?
Junior
Listening for Repeatable, reviewed, versioned changes and less drift.
Follow-up Ask what goes wrong when someone changes things by hand anyway.
-
14
How do you make a change with Terraform safely? What do you check before you apply it?
Junior, Mid
Listening for Reads the plan carefully, watches for destroys, uses review and state locking.
Follow-up Ask about a plan that surprised them.
-
15
Two engineers changed the same Terraform resources and now the plan looks scary. How would you sort it out?
Mid, Senior
Listening for Stops further applies, compares state with reality, fixes carefully with review.
Follow-up Ask how they would prevent it next time.
Distributed systems
-
16
One service calls another and sometimes times out. What could go wrong if you just add retries?
Junior, Mid
Listening for Knows retry storms, backoff, timeouts and duplicate requests.
Follow-up Ask how they would make a retried request safe.
-
17
Requests are slow only sometimes, and each one goes through five services. How would you find the slow part?
Mid, Senior
Listening for Uses tracing and percentiles rather than averages, looks for shared dependencies.
Follow-up Ask what they would do without tracing in place.
-
18
Our main cloud region goes down. How would you design things so we're back within an hour?
Senior
Listening for Backups and data copies in another region, tested failover, infrastructure as code, clear decision owner.
Follow-up Ask how often they would test it.
Incident response
-
19
Tell me about an incident you were part of. What did you do, and what did you learn?
Junior
Listening for Clear about their own part and the technical cause.
Follow-up Ask what they would check first if it happened again.
-
20
You're leading an incident and the site is down for everyone. What do you do in the first fifteen minutes?
Mid, Senior
Listening for Takes a clear role, gets the right people, focuses on restoring service and updates.
Follow-up Ask how they avoid everyone debugging the same thing.
-
21
How do you run a postmortem so the same problem doesn't come back, without blaming people?
Senior
Listening for Blameless facts, real root causes, few action items with owners that get done.
Follow-up Ask how they make sure action items are finished.
Go
-
22
When would you write an internal tool in Go instead of Python, or the other way round?
Mid, Senior
Listening for Practical reasons like distribution, speed, team skills and libraries.
Follow-up Ask about a tool where they chose wrong.
Automation
-
23
Tell me about a task you automated with a script. How did you make sure it was safe to run twice?
Junior, Mid
Listening for Thinks about running again, failures halfway and dry runs.
Follow-up Ask how they tested it before running it in production.
-
24
How do you decide which manual ops work to automate first?
Senior
Listening for Measures time spent and risk, picks high-value work, weighs the cost of keeping automation alive.
Follow-up Ask about something they decided not to automate.
Linux
-
25
A disk on a server is full. How would you find what filled it and fix it safely?
Junior
Listening for Finds the big files before deleting, thinks about logs and rotation.
Follow-up Ask how they would stop it filling up again.
-
26
A Linux server is slow, but nothing in the logs looks wrong. What would you check first?
All levels
Listening for Looks at CPU, memory, disk and network in a clear order.
Follow-up Ask what they would do if every number looks normal.
Networking
-
27
After a network change, a service can't reach the database. How would you debug it step by step?
All levels
Listening for Works layer by layer: DNS, routes, security groups, ports, credentials.
Follow-up Ask what they would check first and why.
Cloud security
-
28
Where should passwords and API keys live for an app in production, and why not in the code?
All levels
Listening for Secret manager, access control, rotation, nothing in git or images.
Follow-up Ask what they would do if a key was pushed to a public repo.
On call
-
29
You're on call for the first time and get paged for a system you don't know well. What do you do?
Junior
Listening for Checks the runbook and dashboards, escalates early instead of guessing.
Follow-up Ask when exactly they would wake someone up.
Developer tooling
-
30
Have you built a small tool or script for your team, in Go, Python or shell? What problem did it solve?
Junior
Listening for A real problem, a simple solution, and people using it.
Follow-up Ask how they shared it and kept it working.
-
31
Developers wait two days for a new environment. How would you build a self-service platform for them?
Senior
Listening for Starts from developer needs, safe defaults and templates, limits on cost.
Follow-up Ask how they would decide what not to build.
Iam
-
32
A developer asks for admin access to production just for today. What do you do?
Junior, Mid
Listening for Finds out the real need and gives the smallest access for a limited time.
Follow-up Ask how they would handle it if a manager insisted.
-
33
How would you design cloud access for fifty engineers so it stays safe but doesn't slow them down?
Senior
Listening for Roles and groups, single sign-on, short-lived access, audit, and break-glass.
Follow-up Ask how they would roll it out to a team used to broad access.
Docker
-
34
What do you think about when you write a Dockerfile for a service?
Junior
Listening for Small base image, layer caching, no secrets in the image, non-root user.
Follow-up Ask how they would make a slow image build faster.