The Kube*Overcommit alerts detect when the current resource requests wouldn't be honored if one node goes down.
In a single node environment, these alerts will always fire because by definition running workload couldn't be rescheduled in case of node failure.