Why I Prefer Good Testing Over Excellent Testing
summary
Stories name the failure modes and the fixes: a monitoring user skewed the usage report, an actual port that was closed in one datacenter, an S3 outage breaks features with no fallback, a new event-messaging service that accidentally pings once per second (not ten minutes) wasting money, and a rollout managed by LaunchDarkly feature flags allowing one‑click rollback. The talk states concrete numbers, monitoring and cost telemetry used, and shows what was built and how rollouts and rollbacks actually worked in production.
AI-written summary · about
chapters (8)
- 0:00Start
- 2:10Monitoring user REALLY loves the home page
- 3:54The port that was closed for no reason
- 5:32Yeah, but S3 will never go down
- 7:33The off by 10,000 error
- 11:33Story 5: The calmest production failure ever
- 12:431. Do good testing
- 13:172. Plan for failure
related talks
28mProduction Testing Through Monitoring
01h 02mThe Art & Joy of Testing in Production
32mI Test In Production
24mHow to solve intermittent issues, aka Random Failures
50mDevOps Project Failures - Lessons Learned and Strategies for Recovery
56mTesting at scale
24mThe more we test, the worse?
30mReducing Chaos in Production: Testing vs. Monitoring