(helm/v1alpha1): Prometheus integration is not working well when enabled in the Chart

[camilamacedo86]

What broke? What's expected?

We need to add in our GitHub action to test the Helm Chart:
https://github.com/kubernetes-sigs/kubebuilder/pull/4377/files#diff-c93a79ad917924bb6a8d743a2c2b36706eeb045ed2e6b42b67b7928da9e7726aR108-R117

Then, we need to:

• We need to ensure that we replace prometheus: false with true in the values.yaml for the CI
• We need to fix the issue

Run helm install my-release ./dist/chart --create-namespace --namespace test-project-system
Error: INSTALLATION FAILED: unable to build kubernetes objects from release manifest: resource mapping not found for name: "test-project-controller-manager-metrics-monitor" namespace: "test-project-system" from "": no matches for kind "ServiceMonitor" in version "monitoring.coreos.com/v1"
ensure CRDs are installed first
Error: Process completed with exit code 1.

Reproducing this issue

No response

KubeBuilder (CLI) Version

master

PROJECT version

No response

Plugin versions

No response

Other versions

c/c @monteiro-renato

Extra Labels

No response

[monteiro-renato]

Hey 👋
This is "expected" if the CRD hasn't been deployed to the cluster first.
That has been mentioned in the design doc as well.

[monteiro-renato]

You could do a similar thing to what is being done in the e2e tests, which is to install the prometheus operator bundle beforehand.

[monteiro-renato]

I can add the missing bits to the Github Action.
Simply installing the CRDs as a dependency should fix the problem.
There's an helm for the CRDs, no need to install the bundle.

[monteiro-renato]

/assign

[monteiro-renato]

Ready for review!