GrafanaDashboard folder handling

[marcantoine-bibeau]

Describe the bug
GrafanaDashboard currently have a spec.folder that is the folder name to be used or created. It does NOT "connect" to the GrafanaFolder CR

Version
v5.8.1

To Reproduce

• Create a new GrafanaDashboard with a "folder name" when there's NO GrafanaFolder
• Dashboard is created inside a new "folder name" Folder

Expected behavior
It should follow the same strategy that GrafanaAlertRuleGroup follow using the folderRef instead.
Otherwise GrafanaAlertRuleGroup should follow GrafanaDashboard folder strategy

Runtime (please complete the following information):

• OS: Linux
• Grafana Operator Version v5.8.1
• Environment: Kubernetes
• Deployment type: running the operator on Kubernetes with Grafana Cloud

[theSuess]

Thanks for pointing this out. We'll deprecate the folder field of the GrafanaDashboard resource and introduce the same fields as in GrafanaAlertRuleGroup (folderRef and folderUID).

Removing the folder field entirely though is not something we can easily do without breaking changes. Since we had some other breaking changes proposed now, we have opened a new milestone for promoting the CRD from v1betav1 to v1. I've added this issue to that list now.

[ahmada-liatrio]

I think that there's some flexibility to be gained having the GrafanaAlertRuleGroup use the GrafanaDashboard folder strategy instead of FolderRef. Makes it easier to use existing folders that may not be captured in code through the operator.

I definitely think there should be a standard strategy they both default to.

[mangkoran]

I also encountered related issue. When I tried to use folderRef in GrafanaDashboard, I got error .spec.folderRef: field not declared in schema. I'm on grafana-operator v5.8.1. CMIIW if this has been fixed in recent version.

[theSuess]

@mangkoran if using helm to install the operator, you'll need to perform additonal steps to also update the resource definitions: https://grafana.github.io/grafana-operator/docs/installation/helm/#upgrading

[theSuess]

The issue in general has been resolved through #1601