What would you like to be cleaned:
Under https://kueue.sigs.k8s.io/docs/concepts/preemption/#preemption-strategies we say that LessThanOrEqualToFinalShare: Only preempt a Workload if the share of the preempting ClusterQueue with the preemptor Workload ..., while in https://github.com/kubernetes-sigs/kueue/tree/main/keps/1714-fair-sharing it says [S2-a] value of AlmostLCA(y,x) without z is still higher (or equal) than value of AlmostLCA(x,y) with admitted workload w.
AlmostLCA is not the same as the preempting ClusterQueue if the hierarchy has > 2 levels.
We should adjust the documentation to be consistent with the KEP and the code in the case of hierarchical cohorts.
Why is this needed:
What would you like to be cleaned:
Under https://kueue.sigs.k8s.io/docs/concepts/preemption/#preemption-strategies we say that
LessThanOrEqualToFinalShare: Only preempt a Workload if the share of the preempting ClusterQueue with the preemptor Workload ..., while in https://github.com/kubernetes-sigs/kueue/tree/main/keps/1714-fair-sharing it says[S2-a] value of AlmostLCA(y,x) without z is still higher (or equal) than value of AlmostLCA(x,y) with admitted workload w.AlmostLCA is not the same as the preempting ClusterQueue if the hierarchy has > 2 levels.
We should adjust the documentation to be consistent with the KEP and the code in the case of hierarchical cohorts.
Why is this needed: