Most Asked SRE Interview Questions for Freshers (2027)

60 most asked SRE interview questions for freshers: SLIs, SLOs, error budgets, toil, plus the Linux, DNS, HTTP and TCP basics fresher rounds test.

Sep 17, 2025 - 15:54
Nov 8, 2026 - 02:30
 0  140
Most Asked SRE Interview Questions for Freshers (2027)

Most asked SRE interview questions for freshers test two things: the SRE vocabulary (SLI, SLO, error budget, toil, postmortem) and the systems basics underneath it, such as Linux processes, memory, disks, DNS, HTTP, TCP and simple scripting. This page gives 60 short answers in that order, plus beginner troubleshooting questions you can practise on a laptop.

It is written for graduates and people switching into SRE from support, testing or system administration. For the full question bank see our 100+ SRE interview questions; for production incidents see scenario-based SRE questions; experienced engineers should use advanced SRE questions. All of them link back to the DevOps interview hub.

Key takeaways

  • Fresher SRE rounds spend as much time on Linux and networking as on SRE theory. Do not skip the fundamentals.
  • Be able to turn an availability target into downtime: 99.9% over 30 days is about 43 minutes.
  • Know the four golden signals from the Google SRE book: latency, traffic, errors and saturation.
  • Know what toil is and that the SRE book caps it at 50% of an SRE's time.
  • Practise a fixed troubleshooting order out loud: what changed, what the user sees, which metric or log confirms it.
  • Have one small project to talk about, such as a monitored web app with an SLO dashboard.

What each section covers

SectionQuestionsFocus
Freshers: SRE basics1 to 15Definitions and simple calculations
Freshers: Linux and networking16 to 35The systems knowledge every SRE round checks
2 to 5 years: tools and practice36 to 48Monitoring, scripting, containers and on-call habits
Senior expectations, explained for juniors49 to 52What leads look for so you can aim for it
Beginner troubleshooting scenarios53 to 60Small problems you can reproduce at home

Freshers: SRE basics

1. What does an SRE do, in one sentence?

An SRE uses software engineering to keep services reliable: measuring reliability with SLOs, automating operational work, responding to incidents, and feeding what was learned back into the product.

2. Where did SRE come from?

Google created the role in 2003 under Ben Treynor Sloss and described it in the book Site Reliability Engineering (2016), which Google publishes free at sre.google.

3. Define SLI with an example.

An SLI is a measured ratio of good events to valid events. Example: the share of HTTP requests in the last 28 days that returned a non-5xx response in under 300 ms.

4. Define SLO with an example.

An SLO is the target for an SLI. Example: 99.9% of requests succeed in under 300 ms over a rolling 28 days.

5. How is an SLA different?

An SLA is a contract with customers that has consequences, often service credits, if it is missed. SLOs are internal targets and are usually set stricter than the SLA so the team acts before a breach.

6. What is an error budget?

It is 100% minus the SLO. With a 99.9% SLO, 0.1% of requests may fail in the window. While budget remains, the team can release freely; when it runs out, reliability work takes priority.

7. How much downtime does 99.9% allow in 30 days?

30 days is 43,200 minutes. 0.1% of that is 43.2 minutes. For 99.99% it is about 4.3 minutes, and for 99% about 7.2 hours.

8. What are the four golden signals?

Latency, traffic, errors and saturation. The SRE book says that if you can measure only four metrics for a user-facing system, measure these.

9. What is toil?

Manual, repetitive, automatable work with no lasting value that grows with the size of the service. Restarting the same service by hand every week is toil. Writing a design document is not.

10. How much toil is acceptable?

The SRE book sets a goal of keeping toil below 50% of each SRE's time so the rest goes to engineering that reduces future toil.

11. What is a postmortem?

A written review after an incident that records impact, timeline, root causes and action items. Google's postmortems are blameless: they focus on systems and processes, not on who made a mistake.

12. What is MTTR?

Mean time to restore (or recover): the average time from the start of an incident until service is back. Lower is better. Fast rollback and good alerts reduce it.

13. What is a runbook?

Step-by-step instructions for a known alert or task, written so an on-call engineer can act without guessing. Good runbooks link to dashboards and say when to escalate.

14. What is the difference between monitoring and observability?

Monitoring checks known conditions with predefined metrics and alerts. Observability is how well you can understand new, unexpected problems from the telemetry a system emits: metrics, logs and traces.

15. How is SRE related to DevOps?

DevOps is a set of cultural and technical practices for shared ownership of delivery. SRE is one concrete way to implement them, with specific tools like SLOs and error budgets. Our article are DevOps and SRE the same thing goes further.

Freshers: Linux and networking

16. How do you see which processes use the most CPU?

top or htop sorted by CPU, or ps aux --sort=-%cpu | head. Mention load average and what it means: the number of runnable or waiting tasks, compared with the CPU count.

17. How do you check memory?

free -h. Look at the "available" column, not "free", because Linux uses spare memory for page cache and gives it back when needed.

18. What is the OOM killer?

When the kernel runs out of memory it picks a process to kill, based on an OOM score. dmesg or journalctl -k shows "Out of memory: Killed process". In containers, hitting the memory limit causes an OOMKilled status.

19. A disk is full. Which commands do you use?

df -h to find the filesystem, du -sh /var/* | sort -h to find big directories. If df says full but du does not add up, a deleted file may still be open: lsof +L1 shows it.

20. What is an inode, and how can you run out of them?

An inode stores a file's metadata. Millions of tiny files can use every inode while space remains. df -i shows inode use.

21. How do you read service logs on a systemd host?

journalctl -u nginx --since "1 hour ago" for one unit, journalctl -f to follow, and systemctl status nginx for state plus the last log lines.

22. What do exit codes mean?

0 means success; anything else is failure. 137 usually means the process got SIGKILL (128 + 9), often from the OOM killer. 143 means SIGTERM (128 + 15).

23. What is the difference between SIGTERM and SIGKILL?

SIGTERM asks a process to stop and lets it clean up. SIGKILL ends it immediately and cannot be caught. Kubernetes sends SIGTERM first and SIGKILL after the grace period.

24. What happens when you type a URL into a browser?

DNS resolves the name to an IP, the client opens a TCP connection (or QUIC for HTTP/3), negotiates TLS, sends the HTTP request, and the server or load balancer returns a response. Each step can fail and each has its own tool to test it.

25. How do you test DNS?

dig example.com or dig +short example.com A; dig @8.8.8.8 example.com to query a specific resolver. Check the TTL to know how long old answers may be cached.

26. What are common DNS record types?

A (IPv4), AAAA (IPv6), CNAME (alias), MX (mail), TXT (text, often verification), NS (name servers) and SOA (zone authority).

27. What is the TCP three-way handshake?

SYN from the client, SYN-ACK from the server, ACK from the client. If you see SYNs with no reply, suspect a firewall or a service not listening.

28. TCP or UDP: what is the difference?

TCP is connection-oriented, ordered and retransmits lost data. UDP sends datagrams with no delivery guarantee and less overhead. DNS mostly uses UDP on port 53 and falls back to TCP for large answers.

29. How do you check which ports are listening?

ss -tlnp lists TCP listening sockets with the owning process. ss -s gives a summary of connection states.

30. How do you test an HTTP endpoint from the shell?

curl -sv https://example.com/health shows TLS and headers. curl -o /dev/null -s -w '%{http_code} %{time_total}\n' URL prints status and total time.

31. What do HTTP 4xx and 5xx codes mean?

4xx are client errors (400 bad request, 401 unauthenticated, 403 forbidden, 404 not found, 429 too many requests). 5xx are server errors (500 internal, 502 bad gateway, 503 unavailable, 504 gateway timeout). SLIs usually count 5xx as bad.

32. What is a load balancer?

A component that spreads traffic across backends and removes unhealthy ones using health checks. Layer 4 balancers route on IP and port; layer 7 balancers understand HTTP and can route on paths or headers.

33. What is the difference between latency and throughput?

Latency is how long one request takes. Throughput is how many requests are handled per second. A system can have high throughput and still slow individual requests.

34. Why use percentiles for latency?

An average hides slow requests. p99 tells you the time under which 99% of requests finish, which is closer to what unhappy users experience.

35. What is a TLS certificate and how do you check its expiry?

It binds a domain name to a public key, signed by a CA. openssl s_client -connect example.com:443 -servername example.com </dev/null | openssl x509 -noout -enddate prints the expiry date. Expired certificates cause real outages, so monitor them.

2 to 5 years: tools and practice

36. What does Prometheus do?

It scrapes metrics over HTTP from targets at intervals, stores them as time series, evaluates alerting rules, and sends alerts to Alertmanager, which groups and routes them. Query language: PromQL.

37. Write a PromQL query for request rate.

sum(rate(http_requests_total[5m])) by (service)

rate() works on counters and gives per-second increase over the window. Metric names depend on how the app is instrumented.

38. Write a PromQL query for error ratio.

sum(rate(http_requests_total{code=~"5.."}[5m]))
/
sum(rate(http_requests_total[5m]))

39. What is Grafana used for?

Dashboards and visualisation over data sources such as Prometheus, Loki and many databases. It also has its own alerting.

40. What makes an alert worth paging for?

It needs urgent human action and shows user impact. Anything that can wait until morning should be a ticket, and anything nobody acts on should be deleted.

41. Which scripting language should a fresher SRE know?

Bash for glue and Python (or Go) for anything longer. Interviewers often ask you to parse a log file, count status codes or call an API and handle errors.

42. Count 5xx responses in an access log with shell tools.

awk '$9 ~ /^5/ {count++} END {print count+0}' /var/log/nginx/access.log

This assumes the default combined log format, where field 9 is the status code. Say that assumption aloud.

43. What are liveness and readiness probes?

In Kubernetes, a failing liveness probe restarts the container; a failing readiness probe removes the pod from Service endpoints without restarting it. Mixing them up causes restart loops.

44. What is CrashLoopBackOff?

A pod's container keeps exiting and Kubernetes waits longer before each restart. kubectl describe pod and kubectl logs --previous show why.

45. What is a canary release?

Sending a small share of traffic to the new version first and comparing its errors and latency with the old version before rolling out further.

46. What is the role of an incident commander?

The person who coordinates the response, assigns roles and decides, while others investigate. The SRE book also names operations, communications and planning roles.

47. How should you behave on your first on-call shift?

Read the runbooks before the shift, acknowledge pages quickly, declare an incident early if users are affected, and escalate without embarrassment. Write down what you did as you go.

48. What is infrastructure as code and why do SREs care?

Defining servers and networks in versioned files (Terraform, Ansible) so changes are reviewed, repeatable and easy to roll back. Manual changes are a common incident cause.

Senior expectations, explained for juniors

49. What will a senior interviewer listen for in your answers?

Ordered thinking: impact first, then mitigation, then root cause. Also numbers where you have them and honesty about what you did not know.

50. Why do leads care about error budget policies?

An agreed policy settles release-versus-reliability arguments in advance, so teams do not debate every incident from scratch.

51. Why is "add more monitoring" a weak answer?

More dashboards rarely help. Seniors want the specific signal that would have caught the problem earlier, usually a user-facing SLI.

52. How do you talk about a project you built alone?

Describe the SLO you chose, how you measured it, one failure you caused on purpose, and what you changed afterwards. That shows SRE thinking at small scale.

Beginner troubleshooting scenarios

53. A website returns 502. Where do you start?

502 means the proxy could not get a valid response from the backend. Check whether the backend process is running and listening (systemctl status, ss -tlnp), then the proxy's error log for "connection refused" or timeouts.

54. SSH to a server hangs. What do you check?

DNS resolution of the host, whether port 22 is reachable (nc -vz host 22), security group or firewall rules, and whether the server is overloaded. Use the cloud console's serial output if the network path is fine.

55. A cron job silently stopped running. How do you investigate?

Check the cron logs in the journal, the job's environment (cron has a minimal PATH), file permissions and whether the script exits non-zero. Then add monitoring that alerts when the job has not succeeded for a set time.

56. An app is slow only at 9 a.m. What could it be?

A traffic peak hitting saturation, a scheduled batch job competing for the database, or a cache that expires overnight. Compare latency with traffic and CPU graphs for that hour.

57. A server's load average is 20 on 4 CPUs but CPU use is low. Why?

Load counts tasks waiting on disk I/O too. Check iostat -x or vmstat for high I/O wait. A slow or failing disk or an NFS mount can cause this.

58. A container exits with code 137. What happened?

It was killed with SIGKILL, most often for exceeding its memory limit. Check kubectl describe pod for "OOMKilled" and compare usage with the limit.

59. Users in one office cannot reach the site but others can. Next step?

Ask for dig and curl -v output from that office. Differences in DNS answers, a proxy, or a blocked route usually explain it.

60. You made a change and errors rose. What do you do first?

Roll back, then investigate. Restoring service comes before finding the root cause, and a fast rollback is the cheapest mitigation you have.

A two-week study plan

  1. Days 1 to 3: read the SLO, toil and monitoring chapters of the SRE book.
  2. Days 4 to 7: practise the Linux and networking commands above on a VM until you can explain each output.
  3. Days 8 to 11: run an app with Prometheus and Grafana, define one SLO and build a burn-down panel.
  4. Days 12 to 14: answer the scenario questions out loud with a timer.

More background: the role of SREs in DevOps teams. Our DevOps course and Prometheus and Grafana course cover these labs, in the Kharadi, Pune classroom or live online.

FAQ

What are the most asked SRE interview questions for freshers?

SLI, SLO, SLA, error budgets, toil, golden signals and postmortems, along with Linux and networking basics such as processes, memory, disks, DNS and HTTP status codes.

Can a fresher get an SRE job?

Some companies hire graduates into SRE or production engineering roles, while others hire from support, sysadmin or developer roles. Strong Linux, networking and scripting skills help in either case.

How much downtime does 99.9% availability allow?

About 43 minutes in a 30-day month, or roughly 8.8 hours in a year.

Which programming language should I learn for SRE?

Python is the most common choice for interviews, with Bash for shell tasks. Go is common in infrastructure tooling.

What are the four golden signals?

Latency, traffic, errors and saturation, as defined in Google's SRE book.

Do I need to read the whole Google SRE book?

No. Read the chapters on SLOs, toil, monitoring, incidents and postmortems first.

How is this page different from the 100+ SRE questions?

This page is for beginners and adds the Linux and networking basics fresher rounds test. The 100+ guide covers every level in more depth.

References

What's Your Reaction?

Like Like 0
Dislike Dislike 0
Love Love 0
Funny Funny 0
Angry Angry 0
Sad Sad 0
Wow Wow 0
Mridul I am a passionate technology enthusiast with a strong focus on DevOps, Cloud Computing, and Cybersecurity. Through my blogs at DevOps Training Institute, I aim to simplify complex concepts and share practical insights for learners and professionals. My goal is to empower readers with knowledge, hands-on tips, and industry best practices to stay ahead in the ever-evolving world of DevOps.