Configure GPU node scale up delay #9126
Description
Which component are you using?:
Cluster Autoscaler
/area cluster-autoscaler
Is your feature request designed to solve a problem? If so describe the problem this feature should solve.:
Currently, the autoscaler triggers scales up for the GPU nodes only after 30 seconds whereas for the CPU nodes it triggers after 2 seconds. This is causing too much delay in scale up time for GPU nodes.
Describe the solution you'd like.:
Expose a parameter/flag to configure delay for GPU scale up. This default value for this can be 30 seconds but should have the ability to be overriden.
Describe any alternative solutions you've considered.:
For AWS EKS, I have changed the hardcoded value as per requirement and deployed it inside the cluster. However, for Azure AKS there is no option to use custom docker image for cluster autoscaler.