- Client-go: fixed potential data races retrying requests using a custom `io.Reader` body; with this fix, only requests with no body or with `string` / `[]byte` / `runtime.Object` bodies can be retried ([#113933](https://github.com/kubernetes/kubernetes/pull/113933), [@liggitt](https://github.com/liggitt))

large LIST calls into multiple smaller chunks. A client can specify a limit to the number of results to return. If more results exist, a token is returned that allows the client to continue the previous list call repeatedly until all results are retrieved.  The resulting list is identical to a list call that does not perform chunking, thanks to capabilities provided by etcd3.  This allows the server to use less memory and CPU when very large lists are returned. This feature is gated as APIListChunking...

- When the `RetryGenerateName` feature gate is enabled on the kube-apiserver,
  create requests using generateName are retried automatically by the apiserver when the generated name conflicts with an existing resource name, up to a max limit of 7 retries.

  request optionally sent by the CSI Client from this release onwards. ([#121303](https://github.com/kubernetes/kubernetes/pull/121303), [@humblec](https://github.com/humblec))
- `NodeStageVolume` calls will now be retried if the CSI node driver is not running. ([#120330](https://github.com/kubernetes/kubernetes/pull/120330), [@rohitssingh](https://github.com/rohitssingh))

### Bug or Regression

- Client-go: fixes potential data races retrying requests using a custom io.Reader body; with this fix, only requests with no body or with string / []byte / runtime.Object bodies can be retried ([#113933](https://github.com/kubernetes/kubernetes/pull/113933), [@liggitt](https://github.com/liggitt)) [SIG API Machinery]