Missing Akri-specific Prometheus metrics in the exported metrics endpoint

[fwmarcel]

Missing Akri-specific Prometheus metrics in the exported metrics endpoint

Describe the bug
When scraping Prometheus metrics from Akri, only the default process metrics are exposed (e.g., process_cpu_seconds_total, process_resident_memory_bytes). The Akri-specific metrics described in the documentation, such as akri_instance_count, akri_discovery_response_result, akri_discovery_response_time, and akri_broker_pod_count, are missing.

Output of kubectl get pods,akrii,akric -o wide

pod/akri-agent-daemonset-qz76m                    1/1     Running   0          3m36s   10.42.1.112   rpi-k3s-01   <none>           <none>
pod/akri-agent-daemonset-xw9kt                    1/1     Running   0          145m    10.42.0.190   k3s-server   <none>           <none>
pod/akri-controller-deployment-65db8cd4b6-lvh4q   1/1     Running   0          146m    10.42.1.105   rpi-k3s-01   <none>           <none>
pod/akri-udev-discovery-daemonset-d9dvn           1/1     Running   0          2d1h    10.42.1.41    rpi-k3s-01   <none>           <none>
pod/akri-udev-discovery-daemonset-hkbg8           1/1     Running   0          2d1h    10.42.0.30    k3s-server   <none>           <none>
pod/akri-webhook-configuration-76d4596497-8gsvt   1/1     Running   0          146m    10.42.1.104   rpi-k3s-01   <none>           <none>

NAME                                   CONFIG         SHARED   NODES            AGE
instance.akri.sh/akri-rtl2838-d312d5   akri-rtl2838   false    ["rpi-k3s-01"]   3m10s

NAME                                   CAPACITY   AGE
configuration.akri.sh/akri-3dprinter   1          42h
configuration.akri.sh/akri-rtl2838     1          2d1h

Kubernetes Version:
e.g., Native Kubernetes 1.33.4

To Reproduce
Steps to reproduce the behavior:

1. Install Akri with the Helm command '...'

prometheus:
  enabled: true
udev:
  discovery:
    enabled: true>

3. Access the Prometheus metrics endpoint of the Akri Agent
4. Observe that only the default process metrics are present

Expected behavior
All Akri-specific Prometheus metrics should be exposed, including:

• akri_instance_count (IntGaugeVec, Agent, Configuration/shared)
• akri_discovery_response_result (IntCounterVec, Agent, Discovery Handler name, response result)
• akri_discovery_response_time (HistogramVec, Agent, Configuration)
• akri_broker_pod_count (IntGaugeVec, Controller, Configuration/Node)