Test Automation / Scenario

Nightly Flake Triage

Four repeat offenders, four different root causes, one nightly report. Match each flaky test to its disease and its cure, then set the policy.

Difficulty
Hard
Format
Scenario
Points
200
Estimate
16 min

// MISSION BRIEF

Your Mission

You own the nightly suite: 412 tests, and a flake report that names the same four offenders week after week. Tonight you triage for real: each test gets a root cause and a targeted fix, and the team gets a policy that stops the graveyard from refilling.

The failure signatures differ, and that is the whole point. Read them like symptoms.

Decisions are weighted; the differential diagnosis is most of the score.

// FIRST CONTACT

Battle teaser

First artifact

Flake report (last 30 nightly runs)

checkout_applies_coupon (mid-slide click). Root cause family and fix:

  1. ABackend latency: raise the API timeout
  2. BAnimation race: wait for the transition to finish (stability check) or disable animations in test env, not a click retry loop
  3. CSelector rot: switch the Apply button to XPath
  4. DInfra: move it off runner pool B
Answers, scoring, hints, and the full battle stay sealed.

// SKILL TAGS

test-automationflakinesstriageci-cdtest-strategy