There are several labels used by the Nutanix platform which can vary between instances. If not set as ignore labels on the Cluster Autoscaler, features such as balancing similar node groups will not work predictably.

The Cluster Autoscaler Operator should be updated with the following labels on Nutanix:

* nutanix.com/prism-element-name
* nutanix.com/prism-element-uuid
* nutanix.com/prism-host-name
* nutanix.com/prism-host-uuid

for reference see this code: https://github.com/openshift/cluster-autoscaler-operator/blob/release-4.14/pkg/controller/clusterautoscaler/clusterautoscaler.go#L72-L159

master, 4.14

always

1. create a ClusterAutoscaler CR on Nutanix platform
2. inspect the deployment for the cluster-autoscaler
3. see that it does not have the ignore labels added as command line flags

labels are not added as flags

labels should be added as flags

this should proabably be backported to 4.13 as well since the labels will be applied by the Nutanix CCM