- This PR changes the signature of the `PublishResources` to accept a `resourceslice.DriverResources` parameter instead of a `Resources` parameter. ([#129142](https://github.com/kubernetes/kubernetes/pull/129142), [@googs1025](https://github.com/googs1025)) [SIG Node and Testing]

    *   Kubelet now uses an explicit cm.GetDevicePluginResourceCapacity() function that makes it possible to more accurately determine what resources are inactive and return a more accurate view of available resources.
    *   Extends the device plugin checkpoint data to record registered resources so that we can finish resource removing devices even upon kubelet restarts.

<!-- BEGIN MUNGE: GENERATED_TOC -->

- [v1.17.17](#v11717)
  - [Downloads for v1.17.17](#downloads-for-v11717)
    - [Source Code](#source-code)
    - [Client binaries](#client-binaries)
    - [Server binaries](#server-binaries)
    - [Node binaries](#node-binaries)
  - [Changelog since v1.17.16](#changelog-since-v11716)
  - [Changes by Kind](#changes-by-kind)
    - [Bug or Regression](#bug-or-regression)
  - [Dependencies](#dependencies)

longer available. Kubernetes now only uses the _structured parameters_ model (also alpha) for allocating dynamic resources to Pods.

  if and only if classic DRA was enabled in a cluster, remove all workloads (pods, app deployments, etc. ) which depend on classic DRA and make sure that all PodSchedulingContext resources are gone before upgrading. PodSchedulingContext resources cannot be removed through the apiserver after an upgrade and workloads would not work properly. ([#128003](htt...

### Storage

shutdown. As a consequence of this change the Kubelet now correctly takes into account the resources of running and terminating pods when deciding to accept new pods, since terminating pods are still holding on to those resources. This stricter handling may surface to end users as pod rejections when creating pods that are scheduled to mostly full nodes that have other terminating pods holding the resources the new pods need. The most likely error would be a pod set to `Failed` phase with reason set...

  - The following APIs are no longer served by default:

    - All resources under `apps/v1beta1` and `apps/v1beta2` - use `apps/v1` instead
    - `daemonsets`, `deployments`, `replicasets` resources under `extensions/v1beta1` - use `apps/v1` instead
    - `networkpolicies` resources under `extensions/v1beta1` - use `networking.k8s.io/v1` instead
    - `podsecuritypolicies` resources under `extensions/v1beta1` - use `policy/v1beta1` instead

- API
  - Ingress resources are now available via `networking.k8s.io/v1beta1`. Ingress resources in `extensions/v1beta1` are deprecated and will no longer be served in v1.18. Existing persisted data is available via the new API group/version  ([#74057](https://github.com/kubernetes/kubernetes/pull/74057), [@liggitt](https://github.com/liggitt))

- The AnyVolumeDataSource feature is now beta, and the feature gate is enabled by default. In order to provide user feedback on PVCs with data sources, deployers must install the VolumePopulators CRD and the data-source-validator controller. ([#108736](https://github.com/kubernetes/kubernetes/pull/108736), [@bswartz](https://github.com/bswartz))

    <_comment>Lua source code</_comment>
    <glob pattern="*.lua"/>
    <sub-class-of type="text/plain"/>
  </mime-type>

  <mime-type type="text/x-ml">
    <_comment>ML source code</_comment>
    <glob pattern="*.ml"/>
    <sub-class-of type="text/plain"/>
  </mime-type>

  <mime-type type="text/x-matlab">
    <_comment>Matlab source code</_comment>