If something goes wrong on the target cluster and the controller that handles remote deletions in the root cluster cannot successfully delete the remote machinesets, then we should either set some error status or give up trying to delete the remote machinesets and proceed with the rest of deprovisioning after a given amount of time.