Talor Itzhak 37399c8d4f memorymanager: avoid violating NUMA node memory allocation rule ...

According to https://kubernetes.io/blog/2021/08/11/kubernetes-1-22-feature-memory-manager-moves-to-beta/#single-vs-cross-numa-node-allocation
and to the design introduce in the memory manager KEP: https://github.com/kubernetes/enhancements/tree/master/keps/sig-node/1769-memory-manager
the NUMA node can not have both single and cross NUMA node allocations.

There're cases when the chosen affinity hint does not align with the
rule above but the pod get admitted anyway.

The implications are incosistent admission/rejection behaviors from memory manager side.

In order to fix the issue, we should validate that the affinity hint
that has been chosen is not violating the above rule.

Signed-off-by: Talor Itzhak <titzhak@redhat.com>

2024-07-11 17:04:42 +03:00

35 KiB

Raw Blame History

View Raw