Revisit default alloy scrape configuration

[chancez]

Describe the bug

I'm trying to follow the getting started guide here: https://grafana.com/docs/pyroscope/latest/deploy-kubernetes/helm/

However I'm finding it quite challenging to actually get Alloy to scrape Grafana itself.

There's a few issues in the docs, but I've managed to work around this. The main one being the annotations being used are pyroscope.grafana.com but the default config is expecting profiles.grafana.com along with the profile type being specified. However, even after fixing the pod annotations, it's still not working.

I believe the issue is because of the following line that's added to the Alloy scrape config: https://github.com/grafana/pyroscope/blob/main/operations/pyroscope/helm/pyroscope/templates/configmap-alloy.yaml#L86-L90

It seems to be the case that the Alloy config is expecting a container port to be exposed that matches the port configured in the scrape annotations. The problem is, I don't see a way to configure additional ports ion the Grafana pod via the Grafana helm chart: https://github.com/grafana/helm-charts/blob/f2ee54ada5fc925f224dc38986b28c237adaeaa5/charts/grafana/templates/_pod.tpl#L1012-L1021

To Reproduce

Steps to reproduce the behavior:

1. Follow the steps in https://grafana.com/docs/pyroscope/latest/deploy-kubernetes/helm/
2. Notice Grafana is not being scraped.

Expected behavior

Grafana profiles can be scraped.

Environment

• Infrastructure:Kubernetes
• Deployment tool: helm

Additional details

I think it's a pretty common occurance the pods do not define the profiling ports in the ports section of their containers. It's often overlooked when thinking about the ports a container exposes, especially since profiling is optional. It probably doesn't make sense to rely on the container ports defined on the pod, and to simply trust the annotation is referencing a port that is actually listening. This is how it worked for Phlare, and it makes things a lot simpler.