Currently, "Kubernetes nodes by HTTP" supports the macro/filter "{$KUBE.NODE.FILTER.LABELS}" and works great to ignore some specific nodes. However, in the template "Kubernetes cluster state by HTTP" which does kubelet discovery, there is no such option, so kubelets from nodes we don't care about still get collected.

Would be great to be able to ignore these by using labels on the nodes the kubelet runs on.