Detecting leaks overview

Detection happens at three levels:

flowchart LR
  local[Local DevTools]
  logs[Logs and CLI metrics]
  prod[Production monitoring]
  local --> logs
  logs --> prod

Level	Best for
Local	Reproducing, heap snapshots, fixing
Logs	Staging soak tests, CI regression
Production	Trend alerts, incident response

This section covers

1. Chrome DevTools Memory
2. performance.memory — browser live metrics
3. Reading memory graphs — sawtooth vs staircase
4. process.memoryUsage() — Node logging
5. Log signatures — OOM text patterns
6. Monitoring dashboards — Grafana-style trends
7. Grafana & Kubernetes — PromQL, pod restarts, limit %, K8s chart shapes

Detect early

The cheapest fix is catching a 5 MB/hour leak in staging — not a 3 AM OOM in production.