Jack Langston

I think you may have misread. The documentation you linked references a Kubernetes component called the "rescheduler" that does not exist, at least in modern kubernetes distributions. This seems to...

Yep! There is that, but the scheduler still is not going to ignore node taints even for `systen-X-critical` priority classes. Ignoring this node taint still does make sense even in...

I appreciate that perspective, but there is no such thing as a "special" taint in Kubernetes. There were in the past proposals to make `CriticalAddonsOnly` have certain behavior with respect...

@stevehipwell You are making the claim that there is an "existing special taint." Kubernetes has specific documentation for all of common taints which you can find [here](https://kubernetes.io/docs/reference/labels-annotations-taints/); the taint is...

From what I am understanding, you are worried about the following sequence of events: (1) EKS adds a temporary taint with `CriticalAddonsOnly` (2) The CA picks that node as the...

I definitely understand and it sucks that EKS and AKS have such different behavior on this front. What about this solution: (1) For future versions, remove this behavior (2) For...

@pierluigilenoci Well I had to try. Ultimately, we just want our configurations to work regardless of where our cluster is hosted!

Our team's core issue is that currently I can neither tell to the CA to respect the taint nor add additional taints to system nodes in AKS.

@stevehipwell This is indeed the solution we aligned on as well.

I agree with all of the above, but I also think the CA is also something that needs to change as it appears to be making assumptions that are really...