How to use priority expander in gke clusters autoscaler

Question

We are using Cluster Autoscaler in GKE, nodes are scaling up/down, but we wanted a specific machine-family to get prioritize for scheduling, if these machine type not available then use others. (By default it is always preferring n2 over c2, but we wanted reverse)

Then we came across Cluster Autoscaler Priority Expanders, but we are unable to add this to our clusters

Also, seems like GKE already using Price Priority Expanders, but we want Priority Expanders or something similar, is it possible?

Following things tried to prefer c2 machine-family

• We had used preferredDuringSchedulingIgnoredDuringExecution (in deployment yaml) to give high weight to the required node(c2), but it prefer the c2 node from available nodes, and not take any part in prioritizing c2 node for autoscaling.
• Tried to update the Autoscaling Profiles on GKE cluster, didn't find custom or priority options, there are only two options as mentioned here
  • balanced -> currently used by the cluster
  • optimize-utilization

#cluster-autoscaler #expander #priority