Behind the Iron Curtain: Getting Metrics From the Browser into Prometheus
summary
Project sends user-facing metrics and JS errors into Prometheus by embedding a small JavaScript library that emits the Prometheus exposition format, wraps React views with a higher-order render tracker, and does client-side pre-aggregation before pushing to a custom aggregating gateway (Pushgateway rejected due to cardinality). Metrics push every 15s, counters reset on push, alerts trigger on rising error rate and render failures, data persists to disk on the gateway, and the implementation is open source and demonstrated working (dev env showed high latency values to investigate).
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:37Failure modes
- 7:39JavaScript
- 11:02High Request Agency
- 11:57Backend
- 12:42Push Gateway
- 13:38Prometheus aggregating
- 14:08Prometheus artifacts
- 16:09Demo
- 20:37Questions
related talks
37mMaking Your Grandparents Happy With Web App Performance Monitoring
13mLightning Talk: A Drop-in System To Accelerate Metrics Observability by 100x Us...
20mHow We are Dealing with Metrics at Scale on GitLab.com
10mThe Tell-Tale Metric: Replatforming with Edgar Allan Poe
36mFine Grained Mesh Metrics for Better Visibility With Native Performance
34mOptimizing Performance in Large-Scale React & Node.js Applications
11mAre They Really Using It? Monitoring Digital Experience To Determine Feature Effect
24mOne Pane to Rule Them All: Uniting the Prometheus Community with O...