- Fixed a scheduler preemption issue where the victim pod was not deleted due to incorrect status patching. This issue occurred when the preemptor and victim pods had different QoS classes in their status, causing the preemption to fail entirely. ([#126644](https://github.com/kubernetes/kubernetes/pull/126644), [@Huang-Wei](https://github.com/Huang-Wei))

- Fixed a bug that the async preemption feature keeps preemptor pods unnecessarily in the queue. ([#133167](https://github.com/kubernetes/kubernetes/pull/133167), [@sanposhiho](https://github.com/sanposhiho)) [SIG Scheduling]

- Fixed a suboptimal scheduler preemption behavior where potential preemption victims were violating Pod Disruption Budgets. ([#128431](https://github.com/kubernetes/kubernetes/pull/128431), [@NoicFank](https://github.com/NoicFank)) [SIG Scheduling]

- Graduated the asynchronous preemption feature in the scheduler to beta. 
  Now the feature flag (SchedulerAsyncPreemption) is enabled by default. ([#130550](https://github.com/kubernetes/kubernetes/pull/130550), [@sanposhiho](https://github.com/sanposhiho)) [SIG Scheduling]

    *   Added
        [`tf.distribute.experimental.PreemptionCheckpointHandler`](https://www.tensorflow.org/api_docs/python/tf/distribute/experimental/PreemptionCheckpointHandler)
        to handle worker preemption/maintenance and cluster-wise consistent
        error reporting for `tf.distribute.MultiWorkerMirroredStrategy`.
        Specifically, for the type of interruption with advance notice, it