This repository has been archived by the owner on May 25, 2023. It is now read-only.
node idle resources not considered in preemption #911
Labels
kind/bug
Categorizes issue or PR as related to a bug.
lifecycle/rotten
Denotes an issue or PR that has aged beyond stale and will be auto-closed.
Is this a BUG REPORT or FEATURE REQUEST?:
/kind bug
What happened:
There is a pod A (memory: 1GB) running on a node (memory: 2GB), then I scheduled a pod B (memory: 2GB). Pod A was not preempted, because sum of victim resources (1GB) is lower than pod B resource requirements.
What you expected to happen:
Pod A being preempted, because idle resources (1GB) plus victims resources (1GB) is enough to fit pod B (2GB).
How to reproduce it (as minimally and precisely as possible):
Modify
TestPreempt
inpkg/scheduler/actions/preempt/preempt_test.go
.Anything else we need to know?:
I do not know if it is intentional, but node idle resources are ignored when making preemption decisions. I would expect that we want to check if resources from preempting victims plus node idle resources are enough to fit a preemptor.
The text was updated successfully, but these errors were encountered: