The Key To Go Efficiency Is Just a Few Prometheus Metrics Away!
summary
Explains how to collect actionable Go runtime metrics with Prometheus client_golang, using the new runtime/metrics (Go 1.16) API to grab all metrics atomically and filter by regex. Calls out ~90 metrics now exposed, with a community recommended ~10-12 metrics for generic production use. Shows concrete signals to watch: GOMAXPROCS, GOGC (now reloadable via runtime.gogc), and GOMEMLIMIT (G limit since Go 1.19, metric added ~1.21), and gives registry options and examples of when GC tuning or memory limits fix real regressions.
AI-written summary · about
chapters (8)
- 0:00Introduction and speaker intro
- 1:06Why monitor Go runtime metrics
- 5:01Collecting runtime metrics with Prometheus
- 9:57Key recommended Go runtime metrics
- 17:05Memory-related metrics deep dive
- 21:16Concurrency and scheduling metrics
- 23:24Summary and call to action
- 24:46Q&A session
related talks
56mThe Key to Go Efficiency is Just a Few Go Runtime Metrics Away!
44mGo Profiling and Observability from Scratch
27mGrand Treatise of Modern Instrumentation and Orchestration
34mGolang: Specific Performance Issues
51mObscure Go Optimisations
31mHow Go is Making us Faster
26mRespecting Memory Limits In Go
53mDebugging Go Applications