Part 1 of 8 · 1 min read
Why this page comes last
In this module you will
- Explain how a load balancer uses a health endpoint to keep traffic away from a bad server.
- Tell apart liveness, local and dependency health checks, and say what each can get wrong.
- Describe the safety rule for a load balancer when every server looks unhealthy.
- Build a watchdog that restarts a degraded process, and avoid restart loops.
- Describe a system's steady state and use it to judge whether the defences in this level work.
Pages 2 to 7 gave the defences. They all depend on one thing: knowing which parts are well. A server that is up but sick must be spotted, kept out of traffic, and ideally fixed without a person at 3 a.m. This page covers that, and then closes the level by tying the pages together. The books used are "Understanding Distributed Systems" (UDS), chapters 7 and 17, and "Designing Data-Intensive Applications" (DDIA), chapter 8. Where a point is common practice and not from the books, the text says so. The numbers are made up to show the idea, unless the text names a book.