Skip to content

Quiz: Debugging Methodology

← Back to quiz index

4 questions

L0 (1 questions)

1. What are the five steps of the scientific method applied to debugging?

Show answer 1. Observe — identify actual symptoms, not assumptions.
2. Hypothesize — generate multiple possible causes.
3. Predict — if hypothesis X is true, what else should be true?
4. Test — check the prediction, changing one variable at a time.
5. Conclude — confirmed or eliminated? Repeat until root cause found.

L1 (1 questions)

1. Explain the divide-and-conquer approach to debugging a request that fails between a client and a database through 6 components.

Show answer Test the midpoint of the request path first. If the midpoint works, the problem is between the client and the midpoint. If it fails, the problem is between the midpoint and the database. Continue bisecting the remaining half. This is binary search applied to infrastructure — for a 6-component pipeline, you need at most 3 tests instead of 6.

L2 (1 questions)

1. A service started failing at 14:00 and a deployment happened at 13:55. How do you determine if the deployment caused the failure vs mere correlation?

Show answer Three tests for causation:
1. Revert the deployment — if the problem goes away, strong evidence of causation.
2. Reproduce in isolation — can you trigger the failure by making only that change in staging?
3. Explain the mechanism — trace step by step from the change to the symptom. Also check for other events at 14:00: certificate expiry, traffic spikes, cronjobs, or upstream deploys.

L3 (1 questions)

1. Explain the Five Whys technique and why stopping at the first 'why' is insufficient.

Show answer Keep asking 'why' until you reach the systemic cause. Example: DB ran out of connections (why?) -> query held connections for 30 min (why?) -> missing index caused full table scan (why?) -> migration dropped and recreated table without index (why?) -> no automated index validation in migration pipeline. The first why gives you the symptom fix (kill query). The fifth why gives the systemic fix (add CI step comparing indexes before/after migrations) that prevents recurrence.