- Sometimes, the scheduler incorrectly placed a pod in the "unschedulable" queue instead of the "backoff" queue. This happened when some plugin previously declared the pod as "unschedulable" and then in a later attempt encounters some other error. Scheduling of that pod then got delayed by up to five minutes, after which periodic flushing moved the pod back into the "active" queue. ([#120334](https://github.com/kubernetes/kubernetes/pull/120334), [@pohly](https://github.com/pohly))...

- Sometimes, the scheduler incorrectly placed a pod in the "unschedulable" queue instead of the "backoff" queue. This happened when some plugin previously declared the pod as "unschedulable" and then in a later attempt encounters some other error. Scheduling of that pod then got delayed by up to five minutes, after which periodic flushing moved the pod back into the "active" queue. ([#120334](https://github.com/kubernetes/kubernetes/pull/120334), [@pohly](https://github.com/pohly))...

- Sometimes, the scheduler incorrectly placed a pod in the "unschedulable" queue instead of the "backoff" queue. This happened when some plugin previously declared the pod as "unschedulable" and then in a later attempt encounters some other error. Scheduling of that pod then got delayed by up to five minutes, after which periodic flushing moved the pod back into the "active" queue. ([#120334](https://github.com/kubernetes/kubernetes/pull/120334), [@pohly](https://github.com/pohly))...

- Fixes the bug in PodTopologySpread that only happens with QHint enabled, 

# CUDA: This config refers to building CUDA op kernels with clang.
common:cuda_clang --config=cuda
common:cuda_clang --@local_config_cuda//:cuda_compiler=clang
common:cuda_clang --copt=-Qunused-arguments
# Select supported compute capabilities (supported graphics cards).
# This is the same as the official TensorFlow builds.
# See https://developer.nvidia.com/cuda-gpus#compute

import org.jspecify.annotations.NonNull;
import org.jspecify.annotations.Nullable;

/**
 * Static utility methods pertaining to {@link Set} instances. Also see this class's counterparts
 * {@link Lists}, {@link Maps} and {@link Queues}.
 *
 * <p>See the Guava User Guide article on <a href=
 * "https://github.com/google/guava/wiki/CollectionUtilitiesExplained#sets">{@code Sets}</a>.
 *
 * @author Kevin Bourrillion
 * @author Jared Levy

- Previously, the scheduling queue didn't notice any extenders' failures, potentially resulting in missed cluster events and Pods rejected by Extenders being stuck in the unschedulable pod pool for up to 5 minutes in the worst-case scenario. Now, the scheduling queue notices extenders' failures and requeues Pods rejected by Extenders appropriately.

  - apiserver_flowcontrol_current_r: R(the time of the last change in state of the queues)
  - apiserver_flowcontrol_dispatch_r: R(the time of the latest request dispatch)
  - apiserver_flowcontrol_latest_s: S(the request last dispatched) = R(when that request starts executing in the virtual world)
  - apiserver_flowcontrol_next_s_bounds: min and max next S among non-empty queues

### Scheduling

- Improved fairness of the scheduling queue by placing pods which are attempted recently behind other pods with the same priority. ([#73700](https://github.com/kubernetes/kubernetes/pull/73700), [@denkensk](https://github.com/denkensk))

* Change default value of deleting-pods-burst to 1 ([#27606](https://github.com/kubernetes/kubernetes/pull/27606), [@gmarek](https://github.com/gmarek))
* MESOS: fix race condition in contrib/mesos/pkg/queue/delay ([#24916](https://github.com/kubernetes/kubernetes/pull/24916), [@jdef](https://github.com/jdef))