## Question Walk me through the hardest production bug you have diagnosed. Explain the symptom, evidence you collected, hypotheses you formed and rejected, the root cause, the fix, and how you proved the problem was resolved. ### Constraints & Assumptions - Choose a bug you investigated directly. - Use a chronological evidence trail rather than revealing the root cause immediately. - Distinguish correlation from proof. - Include containment, verification, and recurrence prevention. ### Clarifying Questions to Ask - Would you like a customer-facing incident or a long-running correctness bug? - How much system architecture should I establish first? - May I discuss sanitized log fields and metrics? ```hint Make hypothesis changes visible For each major hypothesis, state the observation it predicted and the evidence that supported or ruled it out. ``` ### What a Strong Answer Covers - Impact, timeline, and safe immediate containment. - Specific logs, traces, metrics, queries, or experiments. - A narrowing process that avoids random changes. - Root cause at the relevant technical depth. - Fix validation, rollback, monitoring, and a systemic prevention step. ### Follow-up Questions 1. What was your first wrong hypothesis? 2. Which single observation narrowed the search most? 3. How did you know the fix had not merely hidden the symptom?