* Prevent kubelet from setting allocatable < 0 for a resource upon initial creation. ([#46516](https://github.com/kubernetes/kubernetes/pull/46516), [@derekwaynecarr](https://github.com/derekwaynecarr))

* Bug fixes

- DRA: The scheduler plugin prevented abnormal filter runtimes by timing out after 10 seconds. This was configurable via the plugin configuration's `FilterTimeout`. Setting it to zero disabled the timeout and restored the behavior of Kubernetes <= 1.33. ([#132033](https://github.com...

- Fixed use of webhook admission plugins with multi-version custom resources ([#74154](https://github.com/kubernetes/kubernetes/pull/74154), [@mbohlool](https://github.com/mbohlool))

### AWS

- Prevented AWS Network Load Balancer security groups ingress rules to be deleted by ensuring target groups are tagged. ([#73594](https://github.com/kubernetes/kubernetes/pull/73594), [@masterzen](https://github.com/masterzen))

* Fixed an issue which prevented `gcloud` from working on GCE when metadata concealment was enabled. ([#66630](https://github.com/kubernetes/kubernetes/pull/66630), [@dekkagaijin](https://github.com/dekkagaijin))

* kubelet: prevent removal of default labels from Node API objects on startup ([#54073](https://github.com/kubernetes/kubernetes/pull/54073), [@liggitt](https://github.com/liggitt))

* Bug fix: Clusters with GCE feature 'DiskAlphaAPI' enabled were unable to dynamically provision GCE PD volumes. ([#59447](https://github.com/kubernetes/kubernetes/pull/59447), [@verult](https://github.com/verult))
* Fix a regression that prevented using `subPath` volume mounts with secret, configMap, projected, and downwardAPI volumes ([#61080](https://github.com/kubernetes/kubernetes/pull/61080), [@liggitt](https://github.com/liggitt))



# v1.9.4

- Fix bug that prevented informer/reflector callers from unwrapping and catching specific API errors by type. ([#110076](https://github.com/kubernetes/kubernetes/pull/110076), [@karlkfi](https://github.com/karlkfi)) [SIG API Machinery]

- Reverted regression that prevented `client-go` latency metrics to be reported with a template URL to avoid label cardinality. ([#111752](https://github.com/kubernetes/kubernetes/pull/111752), [@aanm](https://github.com/aanm))

  readable for end users and has a maximum size (4k bytes) to prevent abuse.
  Each pod may have up to 12k bytes of termination messages before the contents
  of each will be truncated.
* Do not delete pod objects until all its compute resource footprint has been
  reclaimed.
  - This feature prevents the node and scheduler from oversubscribing resources

- fixes the errors/warnings in fluentd configuration ([#67947](https://github.com/kubernetes/kubernetes/pull/67947), [@saravanan30erd](https://github.com/saravanan30erd))
- Fixed an issue which prevented `gcloud` from working on GCE when metadata concealment was enabled. ([#66630](https://github.com/kubernetes/kubernetes/pull/66630), [@dekkagaijin](https://github.com/dekkagaijin))