([#123611](https://github.com/kubernetes/kubernetes/pull/123611), [@ritazh](https://github.com/ritazh))

- A new `pod_status_sync_duration_seconds` histogram is reported at alpha metrics stability that estimates how long the Kubelet takes to write a pod status change once it is detected. ([#107896](https://github.com/kubernetes/kubernetes/pull/107896), [@smarterclayton](https://github.com/smarterclayton))...

- Add a namespace label for all `apiserver_admission_* metrics`.
  Expand the histogram range to 0-10s for all `apiserver_admission_*_duration_seconds` metrics. ([#101208](https://github.com/kubernetes/kubernetes/pull/101208), [@voutcn](https://github.com/voutcn))

- Client-go: add new histogram metric to record the size of the requests and responses. ([#108296](https://github.com/kubernetes/kubernetes/pull/108296), [@aojea](https://github.com/aojea))

    - `rest_client_certificate_expiration_seconds` a gauge reporting the lifetime of the current client certificate. Reports the time of expiry in seconds since January 1, 1970 UTC.

  - **apiserver_flowcontrol_priority_level_seat_count_samples**: histograms of seats occupied by executing requests (both regular and final-delay phases included), broken down by priority_level; the observations are taken once per millisecond.

*   `tf.keras`:

    *   Changed the TensorBoard tag names produced by the
        `tf.keras.callbacks.TensorBoard` callback, so that summaries logged
        automatically for model weights now include either a `/histogram` or
        `/image` suffix in their tag names, in order to prevent tag name
        collisions across summary types.

*   When running on GPU (with cuDNN version 7.6.3 or

	      * Time duration tracked was of the initial API call, not the entire operation.

	      * Metrics label tuple would have resulted in many independent histograms stored, one for each disk. (Did not aggregate well).