Machine Config Operator is degraded after OpenShift 4.22 upgrade on VMware vSphere with boot image failure-domain mismatch

Issue

After upgrading a VMware vSphere cluster to OpenShift Container Platform 4.22, the Machine Config Operator can become degraded because automatic boot image management cannot reconcile one or more Machine API MachineSets.

The Machine Config Operator reports an error similar to:

providerSpec workspace (...) does not match any vCenter/failure domain in the Infrastructure object

The affected MachineSets can otherwise remain functional, but boot image management cannot associate their vSphere workspace configuration with a failure domain defined in the cluster Infrastructure object.

The condition can also prevent the cluster from being considered upgradeable until the boot image configuration is reconciled.

For clusters that intentionally maintain the RHCOS vSphere boot image/template manually, attempting to disable automatic MachineSet boot image management while boot image skew enforcement is still operating in Automatic mode can fail with an error similar to:

The MachineConfiguration "cluster" is invalid: <nil>: Invalid value: when skew enforcement is in Automatic mode, any MachineAPI MachineSet MachineManager must use selection mode 'All'