Observability meets a Flaky Test
summary
A flaky UI test in a 20‑year monolith running about 300,000 tests per CI run repeatedly fails with "failed to obtain test suites" only on nightly runs; machines are torn down and logs deleted nightly, so investigation requires remote desktop, DBeaver/MySQL, CI timestamps, web console and source code — having hopped between 7 tools reveals an async setup timer running after the test. Centralize raw events, keep data longevity, add tracing and one queryable data lake to speed root cause.
AI-written summary · about
chapters (5)
- 0:00Start
- 2:34Monitoring vs Observability
- 26:42HOPPING means FRICTION
- 28:49If we go back in the conversation
- 30:56Retrospective
related talks
46mFlaky, Fragile, and Forgotten: The Fall of a Test Automation Project
26mKeeping Your Tests Lean
50mFlaky tests
40mFinding Flakiness: The Real Reasons Your Tests Aren't Reliable
47mHow to Un-Flake Flaky Tests- a New Hire's Toolkit
24mHow to solve intermittent issues, aka Random Failures
36mFixing Flaky Tests Like a Detective
28mTest flakiness; methods for dealing with flaky tests