AZ400-Q: handle a flaky test
A test fails intermittently with no code change, causing developers to rerun the entire pipeline until it passes. Release confidence is declining. What should the team do?
- A) Disable all tests permanently.
- B) Mark every failure as successful.
- C) Add unlimited retries and ignore the failure rate.
- D) Measure the flaky test separately, investigate/isolate the root cause, use limited retries only for understood transient faults, and track pipeline health. ✓
Answer: D. Flakiness is an operational quality problem. Limited retries may reduce noise for a known transient dependency, but they must not hide a defective test or product behavior.
Why not the others: Removing or ignoring tests destroys signal. Unlimited retries conceal the reliability problem and consume capacity.