Go provides a profiling tool called pprof. It’s documented at https://pkg.go.dev/net/http/pprof.
To use pprof with Armada, enable the profiling socket with the following config (this should be under applicationConfig if using the helm charts). This config will listen on port 6060 with no auth.
```
profiling:
port: 6060
hostnames:
For the Armada scheduler, the helm chart will make a service and ingress for every pod. These are named armada-scheduler-0-profiling etc.
For other Armada components, the helm chart will make a single service and ingress called armada-<component name>-profiling. Note calls to these may not consistently go to the same pod. Use kubectl port-forward, or scale the deployment to size 1, if you need to consistently target one pod.