Safely deploying a 100K line Envoy YAML configuration to production
summary
Lyft handles Envoy configs that hit 100K lines and avoids outages by running the Envoy binary in validation server mode, using a standalone config load check tool and an in-house router check tool that executes the routing engine against generated unit tests. Automated dead-route/cluster removals prevent 503s, CI blocks invalid or deprecated fields, generated tests push coverage toward 100%, and route rollouts take about 5–10 minutes instead of manual networking-team reviews.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:18Overview
- 0:40Infrastructure
- 2:35Issues
- 6:11Tech Debt
- 8:14Validation
- 9:36Router Check Tool
- 15:37Improvements
- 18:51Questions
- 20:17Deferring deployments
- 21:30Envoy routes
- 22:49Envoy runtime
- 27:32Summary
related talks
19mSafely Deploying a 100K line Envoy YAML Configuration to Production
32mHow Lyft Used Envoy to Rethink Microservice Development
22mThe Trials and Success of Adopting Envoy at Tinder
11mA decade of Envoy: From Lyft's Debugging Nightmare to the AI Infra...
26mBuilding a Dev/Test Loop for a Kubernetes Edge Gateway with Envoy Proxy
09mLightning Talk: Honey, I Broke the Things: Debugging Gray Failures in Production!
33mIntro: Envoy
30mExtending Envoy: A Guide to Custom Extensions with Envoy Gateway