In Kubernetes pod scheduling and load balancing decisions (I.e.: Kubernetes Cluster Autoscaler) are made based on the resource Requests allocated for every container in a Pod.
In circumstances in which best practices for pods requests allocation are not enforced (I.e.: using Resource Quotas), pods could request more resources than they actually ever use causing overcommitment of nodes. This kubectl plugin lists pods that have a high allocation of resource requests but are using only a small percentage of it.
make test
make build
kubectl waste