Week 2 Live Q&A
summary
Live Q&A focuses on concrete reliability practices: Facebook-grade mitigations use rack-level batteries and wait 45 seconds for generators before firing PLS, so services must finish shutdown or add custom handlers within that window; testing stacks use Jest, Cypress, GitHub Actions, and a dedicated integration framework when provisioning or multi-service topology matters; current branch coverage sits ~20–30% with an aim near 75% coverage (not blocked on PRs); service catalogs, spreadsheets for test cases, fuzzing and load tests are recommended and edge cases surface quickly when automating.
AI-written summary · about
chapters (21)
- 0:00Reliability and testing Q&A
- 1:24Opening remarks and first Q&A
- 2:09SLO specificity with microservices
- 2:52Integration testing vs canary deployments
- 3:56Choosing Just testing framework
- 5:42Battery backup time for navigation
- 7:29Server boot issues after outages
- 9:36Test coverage targets and thresholds
- 12:28Test coverage strategies
- 14:59Integration testing framework benefits
- 16:25Defining SLOs over time
- 19:35Data center power backup
- 21:24Behavior-driven development
- 22:27Test management tools
- 25:20Service catalog ownership
- 26:43PLS shutdown flows design
- 27:45Manual and automated testing teams
- 30:55Co-location and PLS with batteries
- 32:21When service catalog is essential
- 33:27Handling edge cases
- 36:40Closing remarks
related talks
01h 03mDeveloping C++ @ Facebook Scale
23mHow to Build and Deploy Services in the Cloud fireside chat
13mSystems @Scale Summer Live Q&A Session #3
57mCoffee-break with EngX Experts
20mFireside Chat with Emily Freeman
01h 05mMicrotesting
01h 15mTaking Control Of Your Test Environment
47mEverything I know about writing quality software