Meta·Performance Engineer·Onsite - Multi Round
- A web server is returning 500s on order submissions and customers are actively impacted. No recent deploys, no traffic spike. Walk through how you'd diagnose this live, starting from first principles.
- You've identified the immediate cause. How do you separate what you do right now to stop customer pain from what you do later to prevent recurrence?
- One host in your fleet has noticeably higher p99 latency compared to the others. How do you figure out if this is a hardware problem versus a software problem?
- A service keeps crash-looping after a log rotation event. What's your triage process?
“This is the whole interview basically.”