[Feature Request] make new-pod-scale-up-delay configurable in autoscaler profile

[zhiweiv]

Per https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/FAQ.md

By default, scale-up is considered up to 10 seconds after pod is marked as unschedulable, and scale-down 10 minutes after a node becomes unneeded. There are multiple flags which can be used to configure these thresholds. For example, in some environments, you may wish to give the k8s scheduler a bit more time to schedule a pod than the CA's scan-interval. One way to do this is by setting --new-pod-scale-up-delay, which causes the CA to ignore unschedulable pods until they are a certain "age", regardless of the scan-interval. If k8s has not scheduled them by the end of that delay, then they may be considered by the CA for a possible scale-up.

That is exactly what we need for job pod pools. It is good to be able configure new-pod-scale-up-delay in AKS autoscaler profile.

[zhiweiv]

@marwanad
Does AKS plan to implement this in near future? Or just added it to backlog?

[marwanad]

@zhiweiv yup it will be added soon. I'll follow-up with an update.

[github-actions]

Action required from @Azure/aks-pm

[TomGeske]

Current ETA for deployment is end of September 2020.

[zhiweiv]

Any update?

[marwanad]

@zhiweiv This should be available with the September API. CLI support + Docs slated for end of month. Note that the acceptable time unit is seconds so valid values are: 10s, 360s, etc.

[marwanad]

This should be now available in CLI 2.12.0, CLI-extensions 0.4.63 and via ARM templates.

Feel free to report back on this thread if you hit issues.

[zhiweiv]

It should work now, I can see newPodScaleUpDelay in autoScalerProfile after update.

Will https://docs.microsoft.com/en-us/azure/aks/cluster-autoscaler#using-the-autoscaler-profile be updated to reflect this change?