feat(consoleplugin): operator installs and configures ConsolePlugin

[andrewazores]

Welcome to Cryostat! 👋

Before contributing, make sure you have:

• Read the contributing guidelines
• Linked a relevant issue which this PR resolves
• Linked any other relevant issues, PR's, or documentation, if any
• Resolved all conflicts, if any
• Rebased your branch PR on top of the latest upstream main branch
• Attached at least one of the following labels to the PR: [chore, ci, docs, feat, fix, test]
• Signed all commits: git commit -S -m "YOUR_COMMIT_MESSAGE"

---

Fixes: #
Related to cryostatio/cryostat-openshift-console-plugin#3

Description of the change:

This change adds allows the users to provide...

Motivation for the change:

This change is helpful because users may want to...

How to manually test:

1. Insert steps here...
2. ...

[andrewazores]

I think we can remove some more pieces here since we aren't using the patcher Job anymore.

[andrewazores]

2025-01-27T18:53:00Z INFO setup detected OpenShift environment
2025-01-27T18:53:00Z INFO setup found cert-manager installation
&{{0xc00014a750 0xc000693ac0} {0xc00014a750 {}} {0xc000689740 0xc0004299f0} 0xc000280e00 0xc0004299f0 0xc000687e30 map[] false}
2025-01-27T18:53:00Z ERROR setup failed to install OpenShift Console Plugin {"error": "the cache is not started, can not read objects"}
main.main
/opt/app-root/src/internal/main.go:188
runtime.main
/usr/lib/golang/src/runtime/proc.go:271

It seems like this happens because the console plugin installation tries to use the Kubernetes client to create resources (like the ConsolePlugin CR), but since this is being done before mgr.Start() things aren't actually initialized yet. It looks like we can either drop down to using a "raw" HTTP client to install the plugin, or else the ConsolePlugin CR should be handled by a controller as part of the reconcile loop rather than as a one-shot operation before the manager starts?

[ebaron]

2025-01-27T18:53:00Z INFO setup detected OpenShift environment
2025-01-27T18:53:00Z INFO setup found cert-manager installation
&{{0xc00014a750 0xc000693ac0} {0xc00014a750 {}} {0xc000689740 0xc0004299f0} 0xc000280e00 0xc0004299f0 0xc000687e30 map[] false}
2025-01-27T18:53:00Z ERROR setup failed to install OpenShift Console Plugin {"error": "the cache is not started, can not read objects"}
main.main
/opt/app-root/src/internal/main.go:188
runtime.main
/usr/lib/golang/src/runtime/proc.go:271

It seems like this happens because the console plugin installation tries to use the Kubernetes client to create resources (like the ConsolePlugin CR), but since this is being done before mgr.Start() things aren't actually initialized yet. It looks like we can either drop down to using a "raw" HTTP client to install the plugin, or else the ConsolePlugin CR should be handled by a controller as part of the reconcile loop rather than as a one-shot operation before the manager starts?

I'll try using the direct (non-cached) Kubernetes client and see if that works.

[ebaron]

That last commit converts the plugin installer to a Runnable, which gets added to the manager in the same way controllers do. This ensures it doesn't interact with the API server until the manager's cache is started.

For the separate issue of OLM naming the ClusterRoleBinding's unpredictably, I found a shared set of OLM-managed labels between the operator deployment and the ClusterRoleBinding. I use this to find the ClusterRoleBindings belonging to our operator and then iterate until we find the one with the proper Service Account subject.

These two things seemed to work well when tested on OpenShift 4.17.

[ebaron]

/build_test

[github-actions]

/build_test completed successfully ✅.
View Actions Run.

[andrewazores]

GitHub won't let me approve this PR because I'm still technically its author 😁

Looks good and appears to work great. Maintainer override coming up.