I have a 3 k8s cluster nodes. Two of the nodes show about 60% of mem and pod space used. The third node is at 82% Ram and 97% Pod use and yet the scheduler is still placing pods on that node? Even though it's showing a red overloaded state. Why would that be?

the Kubernetes scheduler doesn’t go out of its way to balance things, it takes more of an “if I fits, I sits” approach to placing pods.