- Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins).
  The incorrect loop logic might lead to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which is unexpected. ([#122370](https://github.com/kubernetes/kubernetes/pull/122370), [@caohe](https://github.com/caohe)) [SIG Scheduling]

### Other (Cleanup or Flake)

      segment.cleanUp();
    }
  }

  // ConcurrentMap methods

  @Override
  public boolean isEmpty() {
    /*
     * Sum per-segment modCounts to avoid mis-reporting when elements are concurrently added and
     * removed in one segment while checking another, in which case the table was never actually
     * empty at any point. (The sum ensures accuracy up through at least 1<<31 per-segment

- Added metrics for the nftables kube-proxy mode rather than it reporting
  metrics with "iptables" in their names. ([#124557](https://github.com/kubernetes/kubernetes/pull/124557), [@danwinship](https://github.com/danwinship)) [SIG Network and Windows]

* [Kubelet] Check if kubelet is running as uid 0 ([#30466](https://github.com/kubernetes/kubernetes/pull/30466), [@vishh](https://github.com/vishh))
* Fix third party APIResource reporting ([#29724](https://github.com/kubernetes/kubernetes/pull/29724), [@brendandburns](https://github.com/brendandburns))

* Per Volume Inode Accounting ([#35132](https://github.com/kubernetes/kubernetes/pull/35132), [@dashpole](https://github.com/dashpole))
* [AppArmor] Hold bad AppArmor pods in pending rather than rejecting ([#35342](https://github.com/kubernetes/kubernetes/pull/35342), [@timstclair](https://github.com/timstclair))

        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
        automatically saves a checkpoint, exits the program without raising an

- Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins).
  The incorrect loop logic might lead to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which is unexpected. ([#122368](https://github.com/kubernetes/kubernetes/pull/122368), [@caohe](https://github.com/caohe)) [SIG Scheduling]

### Other (Cleanup or Flake)

- Fixed a regression since `1.24` in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins). The incorrect loop logic might have led to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which was unexpected.
   ([#122068](https://github.com/kubernetes/kubernetes/pull/122068), [@caohe](https://github.com/caohe))

- Evicted and other terminated Pods will no longer revert to the `Running` phase. ([#105462](https://github.com/kubernetes/kubernetes/pull/105462), [@ehashman](https://github.com/ehashman))
- Fix `kube-apiserver` metric reporting for the deprecated watch path of `/api/<version>/watch/...`. ([#104161](https://github.com/kubernetes/kubernetes/pull/104161), [@wojtek-t](https://github.com/wojtek-t))

    yield request_state
    legacy_request_state_context_var.reset(contextvar_token)
```

...before this change it would raise an error when resetting the context variable, because the `contextvars` context was different, because of the way it was implemented.