On-callHardoc-g525

Subject Cold startLevel Senior–Staff~30 minCommon in Reliability & on-call interviewsIndustries Technology

Question

Your fleet has been running warm for days — no deploy, no scale event. At 14:00 you flip a feature flag that turns on a new code path in a hot request handler. Immediately, p99 on an already-warm, fully-scaled fleet spikes for ~3-4 minutes and then settles back to normal, with no replica or instance change. CPU briefly spikes during the window. The flag clearly triggered it, but nothing started cold. How do you triage and explain this 'cold start without a cold start,' and what do you do about it?

What a strong answer looks like

Stop the bleeding first (mitigate), then form hypotheses from real signals. Separate root cause from symptom, communicate status as you go, and close with what prevents a repeat.

Learn the concepts

Diagram & narrate the incident

Loading whiteboard…

Run or narrate your approach, then ask the coach.