What problem does this PR solve?:
Previously, the MetalLB handler created custom resources on the workload cluster. These required a webhook to be ready. The webhook required one or more Nodes to be Ready. That required CNI to be deployed.
When CAPI called the MetalLB handler before the CNI handler, this condition would never be satisified. No handlers after the MetalLB would be called.
With this PR, we delegate the creation of MetalLB configuration on the workload cluster to the ClusterResourceSet controller.
What problem does this PR solve?: Previously, the MetalLB handler created custom resources on the workload cluster. These required a webhook to be ready. The webhook required one or more Nodes to be Ready. That required CNI to be deployed.
When CAPI called the MetalLB handler before the CNI handler, this condition would never be satisified. No handlers after the MetalLB would be called.
With this PR, we delegate the creation of MetalLB configuration on the workload cluster to the ClusterResourceSet controller.
Which issue(s) this PR fixes: Fixes #
How Has This Been Tested?:
Deployed a Nutanix cluster.
Special notes for your reviewer: