"kubectl get cluster-api" prints deprecation warnings

[mboersma]

/kind bug

What steps did you take and what happened:

I ran the command kubectl get cluster-api, but it printed several lines of deprecation warnings about AzureManagedCluster and friends, despite those objects not being present in the cluster.

$ k get cluster-api
Warning: AzureManagedCluster and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedCluster and related AzureASOManaged resources instead.
Warning: AzureManagedClusterTemplate and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedClusterTemplate and related AzureASOManaged resources instead.
Warning: AzureManagedControlPlane and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedControlPlane and related AzureASOManaged resources instead.
Warning: AzureManagedControlPlaneTemplate and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedControlPlaneTemplate and related AzureASOManaged resources instead.
Warning: AzureManagedMachinePool and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedMachinePool and related AzureASOManaged resources instead.
Warning: AzureManagedMachinePoolTemplate and the AzureManaged API are deprecated. Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedMachinePoolTemplate and related AzureASOManaged resources instead.
NAME                                                            AGE
clusterresourceset.addons.cluster.x-k8s.io/crs-calico-windows   28m
clusterresourceset.addons.cluster.x-k8s.io/csi-proxy            28m

NAME                                                                       CLUSTER       AGE
kubeadmconfig.bootstrap.cluster.x-k8s.io/default-851-control-plane-xcfm2   default-851   5m48s

NAME                                                                AGE
kubeadmconfigtemplate.bootstrap.cluster.x-k8s.io/default-851-md-0   8m59s

NAME                                   CLUSTERCLASS   PHASE      AGE     VERSION
cluster.cluster.x-k8s.io/default-851                  Deleting   8m59s   

NAME                                                  CLUSTER       REPLICAS   READY   UPDATED   UNAVAILABLE   PHASE     AGE     VERSION
machinedeployment.cluster.x-k8s.io/default-851-md-0   default-851   2          2       2         0             Running   8m59s   v1.32.0

NAME                                                       CLUSTER       NODENAME                          PROVIDERID                                                                                                                                                           PHASE      AGE     VERSION
machine.cluster.x-k8s.io/default-851-control-plane-xcfm2   default-851   default-851-control-plane-xcfm2   azure:///subscriptions/e240e532-1923-4fe0-86da-28abc43fc4c7/resourceGroups/default-851/providers/Microsoft.Compute/virtualMachines/default-851-control-plane-xcfm2   Running    5m48s   v1.32.0
machine.cluster.x-k8s.io/default-851-md-0-hjtr5-7cgk7      default-851   default-851-md-0-hjtr5-7cgk7      azure:///subscriptions/e240e532-1923-4fe0-86da-28abc43fc4c7/resourceGroups/default-851/providers/Microsoft.Compute/virtualMachines/default-851-md-0-hjtr5-7cgk7      Deleting   5m44s   v1.32.0

NAME                                                 CLUSTER       REPLICAS   READY   AVAILABLE   AGE     VERSION
machineset.cluster.x-k8s.io/default-851-md-0-hjtr5   default-851   1          1       1           8m59s   v1.32.0

NAME                                                                          CLUSTER       INITIALIZED   API SERVER AVAILABLE   REPLICAS   READY   UPDATED   UNAVAILABLE   AGE     VERSION
kubeadmcontrolplane.controlplane.cluster.x-k8s.io/default-851-control-plane   default-851   true          true                   1          1       1         0             8m59s   v1.32.0

NAME                                                                       TYPE              AGE
azureclusteridentity.infrastructure.cluster.x-k8s.io/cluster-identity-ci   UserAssignedMSI   24m

NAME                                                       CLUSTER       READY   REASON   AGE
azurecluster.infrastructure.cluster.x-k8s.io/default-851   default-851   True             8m59s

NAME                                                                           READY   SEVERITY   REASON   STATE       AGE
azuremachine.infrastructure.cluster.x-k8s.io/default-851-control-plane-xcfm2   True                        Succeeded   5m48s

NAME                                                                             AGE
azuremachinetemplate.infrastructure.cluster.x-k8s.io/default-851-control-plane   8m59s
azuremachinetemplate.infrastructure.cluster.x-k8s.io/default-851-md-0            8m59s

What did you expect to happen:

Only to be warned about deprecated objects when they are actually present.

Anything else you would like to add:

Environment:

• cluster-api-provider-azure version: main, post v1.21
• Kubernetes version: (use kubectl version):
• OS (e.g. from /etc/os-release):

[nojnhuh]

I think this is by design? Those warnings get printed for any request to those endpoints, including GET via the cluster-api category whether or not any instances of those resources exist. I tend to list each resource explicitly and end up with insane commands like this in my shell history that I reuse:

kubectl get clusters,azureasomanagedclusters,resourcegroups,azureasomanagedcontrolplanes,managedclusters,machinepools,azureasomanagedmachinepools,managedclustersagentpools,publicipprefixes   

[mboersma]

Those warnings get printed for any request to those endpoints

I figured, it's just not particularly user-friendly in this case when I'm not actually using any of the deprecated resources. If there were something like a CAPZ_NO_DEPRECATION_WARNINGS env var, that would also help.

[nojnhuh]

I guess it depends on what you define as "using" the resources and whether or not that includes GET. I think it's defensible to say that a user trying to GET a deprecated resource should receive a warning.

Those messages ultimately come from kubectl based on how the CRD is defined so I'm not sure there's an easy way to toggle those messages:

cluster-api-provider-azure/config/crd/bases/infrastructure.cluster.x-k8s.io_azuremanagedclusters.yaml

Lines 33 to 36 in 275f9ee

	deprecated: true
	deprecationWarning: AzureManagedCluster and the AzureManaged API are deprecated.
	Please migrate to infrastructure.cluster.x-k8s.io/v1beta1 AzureASOManagedCluster
	and related AzureASOManaged resources instead.

I suppose there might be some kustomize magic we can pull to set those fields on the CRDs more dynamically when CAPZ is installed, but overall I'd prefer to err on the side of being more annoying than strictly necessary in this case.