Top 10 SRE Interview Questions with Expert Answers

01. How to change a process’s priority?

Use renice or the top interface. Example: renice 5 -p 1234 sets the niceness to 5 (higher priority). Negative values require root privileges.

In top, press r, enter the PID, and set the new priority.

02. How to keep a variable from a shell script after it exits?

Variables defined in a child shell disappear when the script ends. Run the script with source script.sh (or . script.sh) so it executes in the current shell, preserving variables.

03. Explain TTD, TTR and why they matter.

TTD (Time To Detect) is the time from a failure occurring to its detection. TTR (Time To Resolve) is the time from detection to full remediation. Lower TTD leads to faster response; lower TTR reduces downtime and improves reliability. Both are key SRE metrics.

04. Design a system architecture for a LinkedIn profile page.

Key components: client apps, API Gateway, Profile Service (microservice), database (MySQL or DynamoDB), cache (Redis/Memcached), media service (object storage like S3), search index (Elasticsearch), connections service, feed service, load balancer, CDN. Non‑functional considerations include caching hot profiles, rate limiting, and security/privacy compliance.

05. Design a system architecture for Twitter.

Core components: client side, API Gateway, User Service, Tweet Service, Timeline Service, media service (S3), search service (Elasticsearch), caching layer (Redis/Memcached), load balancer, CDN. Data stores: MySQL for user data, Cassandra for tweets. Timeline generation can use push (pre‑compute) or pull (on‑demand) models; Twitter typically combines both. Include rate limiting, cross‑region replication, eventual consistency, and scalability for massive tweet volume.

06. Common load‑balancing strategies or techniques?

Round Robin – simple sequential distribution.

Least Connections – routes to server with fewest active connections.

IP Hash – uses client IP for session stickiness.

Weighted Round Robin – gives more traffic to higher‑capacity servers.

Random – selects a server at random.

Consistent Hashing – useful for distributed caches.

Health Checks – ensures traffic only goes to healthy instances.

07. Describe a situation where you strongly opposed a proposal and how you handled it.

Example: the team wanted to skip unit testing to meet a deadline. I explained the risks (technical debt, bugs, future delays), cited a past incident where lack of tests caused production issues, and suggested a minimal set of critical‑path tests. The team agreed to implement those tests, preserving quality without missing the deadline.

08. How to fetch JSON from a REST API?

Use

curl -H "Accept: application/json" https://api.example.com/data

or Python’s requests library:

import requests
response = requests.get('https://api.example.com/data')
data = response.json()
print(data)

09. Write code to parse a log file and count error types.

# parse_logs.py
log_counts = {}
with open('application.log', 'r') as file:
    for line in file:
        if 'ERROR' in line:
            error_type = line.split('ERROR')[1].strip().split()[0]
            log_counts[error_type] = log_counts.get(error_type, 0) + 1
# Display the counts
for error, count in log_counts.items():
    print(f"{error}: {count}")

The script opens the log, scans each line for “ERROR”, extracts the error type, tallies occurrences in a dictionary, and prints the results.