healthy on scale to 0

Describe the Feature You Want

from Alon Avrahami in the k8s slack:

I was able to get KEDA to scale down to zero, but Tilt shows us error: Unhealthy for all of the auto-scaling services because its not aware of the number of replicas running.
I’ve found out the pod_readiness = "ignore" option, and we are currently using it, but I’m not sure its the right way, and i would like to know if you have any other suggestions?
Something that will allow Tilt to check the pod health when replicas > 0, and allow ignore when replicas = 0. (

Current Behavior

if there are 0 instances of a pod, tilt treats it as unhealthy

Additional context

I'm not quite sure what the right way to solve this is.

in the general case, there's not an easy way for tilt to tell the difference between "you have 0 replicas because of some catastrophic failure" and "you have 0 replicas because KEDA scaled down". These are also not mutually exclusive cases! (Imagine: KEDA scaled down, but then you pushed a bad image)

maybe the simple solution is to have some sort of "Absent" state, and there's a way to mark the Absent state as a healthy one 🤔

Start by tracing Tilt's pod health handling and the existing pod_readiness = "ignore" option. Clarify how Tilt could distinguish an intentional zero-replica KEDA scale-down from a failure, then define the desired health state and its behavior when replicas are zero versus greater than zero.