Testing in Production
summary
Hands-on techniques and outcomes: phased rollouts via canaries and blue/green with scripted auto rollback, smoke tests scheduled (every few hours) that caught a third‑party returning empty bodies, and keeping smoke suites small (under a dozen, often <6). Tools and patterns include avoiding VCR for prod checks, Chaos Monkey/Simian Army style controlled breakage and DiRT disaster drills, routine DR restores from backups after an accidental schema run, and treating alerts as tests with actionable thresholds and cleanup to leave no trace.
AI-written summary · about
chapters (33)
- 0:00Intro
- 0:47Google Cloud Platform
- 1:19Victory Conditions
- 1:53Intentionality
- 2:29Traditional Tests
- 2:54Blackbox Tests
- 8:08Deployment Techniques
- 8:51Choosing Canary Group
- 9:24Demographic
- 10:27Easy Rollback
- 11:41Variation
- 12:58Session Affinity
- 13:17Separate Data Migrations
- 14:03Basic Experimental Design
- 15:20Make Expectations Clear
- 15:40Reusing Tests
- 16:10Smoke Tests Against Prod
- 18:383rd Party Integrations
- 18:59Leave No Trace
- 19:33Controlled Breakage
- 19:52Response & Recovery
- 21:05Simian Army
- 21:41Penetration Testing
- 22:13Ethical Breakage
- 22:48Disaster Recovery Verification
- 24:24Testing Scripts
- 24:42Implicit Testing
- 25:09Monitoring
- 25:29Tell Us The System Isn't Meeting Expectations
- 27:13Dos and Don'ts
- 27:33Don't DDOS Yourself
- 28:14Do Test Your Seams
- 29:40Do Act Methodically
