Debug an OTel Collector using zpages and pprof extensions

Steps

1. Add both extensions to the extensions block in config.yaml: zpages with endpoint: 0.0.0.0:55679 and pprof with endpoint: 0.0.0.0:1777; then list both under service.extensions so they are activated
2. Access zpages in a browser or via curl at http://localhost:55679/debug/pipelinez to see live per-pipeline stats (spans/sec accepted, refused, dropped); /debug/servicez shows Collector version and uptime; /debug/extensionz lists active extensions
3. Use /debug/tracez (if the Collector is built with trace support) to inspect sample trace data flowing through the Collector without sending it to a backend
4. For CPU profiling, use go tool pprof http://localhost:1777/debug/pprof/profile?seconds=30 to capture a 30-second CPU profile; analyse with go tool pprof -http :8090 to open a web-based flame graph in your browser
5. For memory profiling (heap), run go tool pprof http://localhost:1777/debug/pprof/heap; look for large allocations in the batch processor or receivers that may indicate buffer misconfiguration or a memory leak
6. Restrict both extension endpoints to localhost or an internal network—do not expose port 1777 or 55679 on public interfaces since pprof can leak sensitive runtime information and zpages exposes pipeline internals