targetPort on cortex proxy should be 'http' not 'metrics'

[rdvhoorn]

Bug description

the cortex-proxy-metrics svc targets pods has the following spec:

spec:
  ports:
    - name: metrics
      port: 8080
      protocol: TCP
      targetPort: metrics
  selector:
      app.kubernetes.io/instance: <can be custom> 
      app.kubernetes.io/name: cortex-proxy

The cortex proxy pods do not define this port, and only define the an http port and a health port.

Expected behavior

By default, the cortex proxy svc should target the right port on the pod.

Additional context

For my current setup, i patched the svc, after which it worked fine.

I've installed the cortex proxy as follows:

helm upgrade --install mimir-proxy oci://ghcr.io/projectcapsule/charts/cortex-proxy   --namespace observability   -f cluster/observability/values/tenancy-proxy/mimir-proxy-values.yaml

With the following values:

config:
  backend:
    # Mimir remote_write endpoint is exposed via mimir-gateway.
    # The distributed chart usually supports Prometheus remote_write at /api/v1/push on the gateway.
    url: "http://mimir-gateway.observability.svc.cluster.local/api/v1/push"

  tenant:
    # Route by this label found in incoming timeseries.
    labels:
      - "namespace"

    default: admin

args:
  logLevel: 5

[rdvhoorn]

What i dont get is that the service file in the chart seems to not be the same as the one that is loaded for me. So maybe im installing wrong? However the install instructions in the readme of the cortex proxy dont seem to work for me:

helm upgrade --install mimir-proxy oci://ghcr.io/projectcapsule/charts/cortex-proxy --version 0.4.7

Results in

Error: failed to perform "FetchReference" on source: ghcr.io/projectcapsule/charts/cortex-proxy:0.4.7: not found