I am trying to better monitor some application costs, and wanted to take advantage of the GKE cost allocation feature by labelling my workloads appropriately. This goes fine for our autopilot clusters, but we also have several standard clusters where the pod or workload level labels are not used. I am certain that allocation is enabled, the labels are set a properly, there are less than 50, they are not longer than 63 chars and that each pod has resource requests set.
Any ideas? Should this work with a standard cluster?