* The following deprecated flags were removed from `kube-controller-manager`:

  * `replication-controller-lookup-cache-size`
  * `replicaset-lookup-cache-size`
  * `daemonset-lookup-cache-size`

  Don't use these flags. Using deprecated flags causes the server to print a warning. Using a removed flag causes the server to abort the startup.

* kubelet: resolved hang/timeout issues when running large numbers of pods with unique configmap/secret references by reverting to 1.11 configmap/secret lookup behavior ([#74755](https://github.com/kubernetes/kubernetes/pull/74755), [@liggitt](https://github.com/liggitt))

* Disable proxy to loopback and linklocal ([#71980](https://github.com/kubernetes/kubernetes/pull/71980), [@micahhausler](https://github.com/micahhausler))
* kube-scheduler: restores ability to run without authentication configuration lookup permissions ([#71755](https://github.com/kubernetes/kubernetes/pull/71755), [@liggitt](https://github.com/liggitt))

- Fix performance regression in scheduler caused by frequent metric lookup on critical code path. ([#117617](https://github.com/kubernetes/kubernetes/pull/117617), [@tosi3k](https://github.com/tosi3k)) [SIG Scheduling]

resources, which represents the list of group aliases (e.g. "all") that every resource belongs to. ([#43338](https://github.com/kubernetes/kubernetes/pull/43338), [@fabianofranz](https://github.com/fabianofranz))

* `--service-account-lookup` now defaults to true, requiring the Secret API object containing the token to exist in order for a service account token to be valid. This enables service account tokens to be revoked by deleting the Secret object containing the token. ([#44071]...

- Fixed performance regression in scheduler caused by frequent metric lookup on critical code path. ([#117594](https://github.com/kubernetes/kubernetes/pull/117594), [@tosi3k](https://github.com/tosi3k))

- `client-go`, using log level 9, traces the following events of a HTTP request:
      - DNS lookup
      - TCP dialing
      - TLS handshake
      - Time to get a connection from the pool

and may be called again to continue the loop.
As long as the loop body does not terminate, the "range" clause will continue
to generate iteration values this way for each <code>yield</code> call until
<code>f</code> returns.
If the loop body terminates (such as by a <code>break</code> statement),
<code>yield</code> returns false and must not be called again.
</li>
</ol>

<p>

* Performs validation (when applying for example) against OpenAPI schema rather than Swagger 1.0. ([#51364](https://github.com/kubernetes/kubernetes/pull/51364), [@apelisse](https://github.com/apelisse))
* Make all e2e tests lookup image to use from a centralized place. In that centralized place, add support for multiple platforms. ([#49457](https://github.com/kubernetes/kubernetes/pull/49457), [@mkumatag](https://github.com/mkumatag))

- Fixed performance regression in scheduler caused by frequent metric lookup on critical code path. ([#116441](https://github.com/kubernetes/kubernetes/pull/116441), [@alculquicondor](https://github.com/alculquicondor)) [SIG Scheduling]