Skip to content

profiling.md

Latest commit

 

History

History
34 lines (23 loc) · 656 Bytes

profiling.md

File metadata and controls

34 lines (23 loc) · 656 Bytes
Raw

Profiling NetObserv components

All go components use pprof for profiling.

FLP

In the FlowCollector config, turn on profiling:

spec:
  processor:
    advanced:
      profilePort: 6060

After a pod restarted, port-forward its profiling port to localhost:

kubectl port-forward pods/flowlogs-pipeline-27wl8 6060:6060 -n netobserv

You can verify that the dashboard is present on http://localhost:6060/debug/pprof/.

Start profiling, e.g. for CPU:

curl "http://localhost:6060/debug/pprof/profile?seconds=10" > profile10s_1

Use the web UI to analyse the data:

go tool pprof -http=:8080 profile10s_1