use case: I have had to move my lab to another gke cluster due to punishing my little Intel NUC running harvester and running into the same sort of issue deploying minio as some of my nodes due not meet the CPU requested. So I figure that would be a use case for either HPA or VPA. My immediate resolution would be to deploy other machine type nodes with higher CPU into the pool and then draining and decomming the older ones.... but playing with auto-scaling seems 'fun' 😉