I Don’t Always Test My Code, but When I Do It’s in Production
summary
Deployment pipeline shows a concrete claim: replacing manual canaries with automated canary analysis (Spinnaker + Kayenta) finds regressions that human checks miss. The demo pulls production images, creates a control and a canary, streams metrics from Prometheus, runs a Netflix-style judge, and scores across weighted metric groups. A real bug had caused ~30% of registrations to drop while CPU looked better; the analysis flags lower median latency and fails, cleans up, and prevents promotion.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:08Production deployment disaster
- 3:30Lessons from the failure
- 4:54Defining risk in deployments
- 6:19Reducing failure impact
- 7:44Spinnaker canary orchestration
- 10:32Canary analysis setup
- 12:17Configuring canary metrics
- 15:26Running the canary analysis
- 18:16Demo results and cleanup
- 19:42Conclusion and Q&A
