Test Automation. Worst Practices
summary
Juno replaced one-size-fits-all testing with per-service strategies: services are split by role, database setup is centralized in a library that generates constrained test rows, and transport and message schemas live in a contract library while transports have their own client library. The team measures time-to-start-testing (about one day) and built a plugin that flags slow green tests, finding ~800 tests over 2s and ~200 real issues; they also use rule-based dictionary comparisons and randomized but class-equivalent data generation to reduce noise and make failures actionable.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 1:24Universal test framework attempt
- 3:16Microservice testing evolution
- 5:44Test type separation
- 7:33Test design errors impact
- 10:02Excess test maintenance
- 14:34Testing efficiency metrics
- 19:09Test noise impact
- 22:22Testing goals and naming
- 26:42Logging and test data
- 29:09Outdated result changes
- 33:46Fixing test problems
- 36:57Test runtime analysis
- 38:48Tools for test issues
- 41:18Q&A
- 43:14Team testing organization
- 45:59Integration testing isolation
- 47:32Frontend testing
- 48:14Framework maintenance
- 51:14Dev and test sync
- 53:21Conclusion
related talks
45mBuilding a successful test automation architecture in a microservices environment
35mTest automation guide: How to move away from chain tests
19mAutomatic Testing for Services at Scale
01h 05mMicrotesting
57mAdvanced Testing Techniques: Tips from the trenches
47mThe Valley of Success
46mTesting the untestable - patterns and use cases analysis
22mTest Automation and Beyond: Implementing Continuous Testing Effectively