What you would like to be added?
When a node is full, and an in-place resize is performed for a pod within the node, KAI scheduler should pre-empt/reclaim other pods in the node to allow resizing to complete
Why is this needed?
So that we can automatically scale resource requests of important workloads to make room for other less important workloads while also protecting the important workload to ensure that if it needs to scale up it can scale up.
What you would like to be added?
When a node is full, and an in-place resize is performed for a pod within the node, KAI scheduler should pre-empt/reclaim other pods in the node to allow resizing to complete
Why is this needed?
So that we can automatically scale resource requests of important workloads to make room for other less important workloads while also protecting the important workload to ensure that if it needs to scale up it can scale up.