After upgrade to 4.12.28 from 4.11.46 noticed that pods are not getting scheduled evenly across nodes.
Nodes with only 400m - 600m cpu and less then 5% memory commitment will not have pods scheduled to them.
instead nodes with 1000-1500m of commitment will continue to have the majority of pods schedule to those nodes.

lowNodeUtilization has been applied as the scheduler profile but that did not result in any change to how the pods are scheduled.