Basically, OpenShift(Kubernetes) controller determines how many pods can be scheduled on the worker nodes across the cluster through configured requests resources on new pods. But if you have not configured any requests resources on the new pods, then theoretically you can schedule the pod replicas until all worker node allow to run the pod container processes well. It means Kubernetes can not determine how many pod can be scheduled appropriately. So you should configure resources requests(cpu and memory) for best performance and using resources efficiently through the performance test on your application.