Kubernets reduce the number of pods

Question

I have a set of Pods commands that can take up to a couple of seconds. There is a process that tracks an open request and which Pod works. I would like to use this information when scaling the pods - either by indicating which pods to try to leave, or by indicating which jets will be closed. Is it possible to specify this type of information when changing # replicas, for example. I want X replicas, try not to kill my lengthy tasks on shavings A, B, C?

Answer 1

I was looking for a solution for myself, and I also can not find it out of the box. however, a workaround may occur (I would like it if you could check and confirm)

:
 1.
 2. X
 3. X

Answer 2

. , ; "", .

, , - , . , ( ), .