Metrics svc port mismatch on the release/installation file

[gkarthiks]

The target port on the metrics service should be 8080 or metrics instead of https as shown below.

  ports:
  - name: https
    port: 8443
    protocol: TCP
    targetPort: metrics

Since the k8s manifest has the targetPort: https the metrics are not scraped by the Prometheus as there isn't any port on that name specified in the service.

[JohannesLamberts]

@gkarthiks I've just run into this issue causing the service not to become operational: "* the Kubernetes API server reported that "hnc-system/hnc-controller-manager-metrics-service" failed to fully initialize or become live: 'hnc-controller-manager-metrics-service' timed out waiting to be Ready".

The issue you described persists in https://github.com/kubernetes-sigs/hierarchical-namespaces/releases/download/v1.0.0/default.yaml.

This causes our deployment to fail unless we patch the targetPort.

The linked merge-request was closed without further comments.

Should this issue be re-opened, or should I open a new one?

[adrianludwin]

Maybe re-open a new one? I never saw the original issue since it was opened and closed while I was off for Christmas, and I don't really understand the problem. But I'd be grateful if you wanted to patch this into v1.0.1 and can explain what the change does (none of our default tests have Prometheus installed, which may be why we're not seeing this).

[gkarthiks]

hi @JohannesLamberts , If my memory serves me well, I think I closed that after looking at the Kustomization overlays sometime back. But happy to reopen the PR.