Bug: When using PodToDelete along with decreasing replicas, the priority should consider PodToDelete list

[wu8685]

Minimal reproduce step

Scale in 2 pods, along with more than 2, let's say 3 podToDelete.

What did you expect to see?

2 Pods from podToDelete list should be delete with scale-in label, and 1 Pod should be recreated.

What did you see instead

The deletion priority may not consider PodToDelete list.
https://github.com/KusionStack/kuperator/blob/v0.5.3/pkg/controllers/collaset/synccontrol/sync_control.go#L403

What is your KusionStack components and its version?

0.5.0

[ColdsteelRail]

In fact, it works as expected now, as following:
(1) for pods in podToDelete list, we mark the podWrapper.toDelete=true, otherwise false:

kuperator/pkg/controllers/collaset/synccontrol/sync_control.go

Line 173 in c224fb2

ToDelete: toDelete,

(2) then we sort and assign toDelete=true pods with highest priority in getPodsToDelete(),:

kuperator/pkg/controllers/collaset/synccontrol/scale.go

Line 66 in c224fb2

if l.ToDelete != r.ToDelete {

As a result:

• for pods in podToDelete list but out of replicas, they will be scaled in
• for pod in podToDelete list, and in range of replicas, they will be recreate d