Pod scraping requires annotations despite explicit target configuration

[wardbekker]

Current Behavior

When attempting to scrape profiles from a pod using explicit target configuration in Helm values, the scraping fails unless specific Grafana profile annotations are present on the target pod.

Expected Behavior

Profile scraping should work based solely on the explicit target configuration in Helm values, without requiring additional pod annotations.

Configuration

Currently using this Helm values configuration:

profiling:
  enabled: true
  destinations: ["pyroscope"]
  scrape:
    enabled: true
    targets:
      - "__address__": "demo-bank.default.svc.cluster.local:8080"
        "service_name": "demo-bank"

The following annotations are currently required for scraping to work:

profiles.grafana.com/cpu.scrape: "true"
profiles.grafana.com/cpu.port_name: "pprof"
profiles.grafana.com/godeltaprof_memory.scrape: "true"
profiles.grafana.com/godeltaprof_memory.port_name: "pprof"
profiles.grafana.com/godeltaprof_mutex.scrape: "true"
profiles.grafana.com/godeltaprof_mutex.port_name: "pprof"
profiles.grafana.com/godeltaprof_block.scrape: "true"
profiles.grafana.com/godeltaprof_block.port_name: "pprof"
profiles.grafana.com/goroutine.scrape: "true"
profiles.grafana.com/goroutine.port_name: "pprof"
profiles.grafana.com/memory.scrape: "true"
profiles.grafana.com/memory.port_name: "pprof"
profiles.grafana.com/mutex.scrape: "true"
profiles.grafana.com/mutex.port_name: "pprof"
profiles.grafana.com/block.scrape: "true"
profiles.grafana.com/block.port_name: "pprof"

Impact

This creates unnecessary configuration overhead and makes the explicit target configuration in Helm values ineffective on its own.

Additional Context

The explicit target configuration in Helm values should take precedence over or eliminate the need for pod annotations, as it provides all necessary information for scraping.