Scheduling on basis of Scalers using KEDA

[xi2817-aajgaonkar]

instead of just time if we can schedule our pods on basis of metrics like HTTP request on ingress, it would be better

We can use keda for it
https://medium.com/backstagewitharchitects/how-autoscaling-works-in-kubernetes-why-you-need-to-start-using-keda-b601b483d355

[xi2817-aajgaonkar]

we can discuss this feature

[davidebianchi]

Hi @xi2817-aajgaonkar, it would be interesting but not so easy to scale to 0 based on incoming HTTP request (and to recover the number of replicas at the first incoming call), do you have some ideas?

This operator is more intended as development support, so for the wake up we could also lose the first API calls maybe.

[jwcesign]

if want to scale to 0, check this: https://github.com/knative/serving/blob/main/docs/scaling/images/scale-from-0.png
Maybe can use a middle component to cache it, and when it's not zero, route to the service.