What’s my App *Really* Doing in Production?
summary
Shows how to instrument Rails apps with Active Support Notifications and log correlated metrics so cache hits and latencies appear together, demonstrates using a per-thread mattr to scope measurements, warns about global TracePoint and suggests using the C API version as a single-thread probe with TracePoint, recounts a Mongrel memory-cache thrash from a TV traffic spike and an API where JSON serialization, not DB, explains surprising slow requests, and recommends sampling and async reporting.
AI-written summary · about
chapters (5)
- 0:00Start
- 25:04Avoid side effects Avoid behavior changes, not just state changes
- 27:21Measure Results (responses, error rates)
- 28:23Internal processes (intermediate results, rendering)
- 31:07Start measuring!
related talks
36mIt's not in Production Unless it's Monitored
30mRails Performance monitoring: A primer for developers
37mWhat’s happening in your Rails app? Introduction to Introspection features of Ruby
34mMetric Driven Development with Ruby on Rails
43mReal World Ruby Performance at Scale
48mOn the Edge of Rails Performance
37mThe Future of Ruby Performance Tooling
29mHow to Performance