Florent Beaurain: Optimizing Rails Tests at Doctolib Scale

On Rails

فصول

1. Florent Beaurain: Optimizing Rails Tests at Doctolib Scale (00:00:00)

2. Doctolib’s CI suite runs 84,000+ tests per commit, using over 130 CPU hours (00:04:56)

3. Scaling test infrastructure with 45-minute pipelines, test selection, and parallel servers (00:07:38)

4. Improving local dev experience by letting engineers run isolated engine tests (00:10:31)

5. Database resets and factories identified as key bottlenecks for test performance (00:12:47)

6. Switching to transactional tests and revisiting fixtures to align with Rails defaults (00:14:23)

7. Dropping one engine’s test time from 7 minutes to under 1 minute (00:18:44)

8. Migrating 90% of the codebase to a faster testing framework in three months (00:25:50)

9. Using Packwerk to modularize the monolith—why zero dependency engines are a myth (00:31:03)

10. Leveraging AI to automate cleanup tasks and support onboarding (00:36:14)

11. Hitting AWS Aurora scaling limits with 10 Postgres writers and 15 readers each (00:43:20)

12. Avoiding downtime with multi-step database migrations and rollback strategies (00:50:15)

13. Staying current with Rails via dual-booting, CI-driven development, and upstream tracking (00:52:11)

14. Advice for smaller teams upgrading Rails: read the source code and start small (00:56:37)

15. Managing 20,000 end-to-end tests with retry logic and Capybara Lockstep (01:06:23)

16. Using internal CLI tool (dctl) to streamline local setup and staging access (01:10:35)