- You can now have different operation names for different storage operations. This still prevents two operations on same volume from happening concurrently but if the operation changes, it resets the exponential backoff.
([#75213](https://github.com/kubernetes/kubernetes/pull/75213), [@gnufied](https://github.com/gnufied))

   ([#122558](https://github.com/kubernetes/kubernetes/pull/122558), [@linxiulei](https://github.com/linxiulei))
- Fixed CEL estimated cost for expressions that perform operations on the result of `map()` operations (e.g., `.map(...).exists(...)` ) to have the correct estimated cost instead of an unbounded cost.
   ([#123562](https://github.com/kubernetes/kubernetes/pull/123562), [@jpbetz](https://github.com/jpbetz))

* Azure cloudprovider: Reduce the polling delay for all azure clients to 5 seconds. This should speed up some operations at the cost of some additional quota. ([#43699](https://github.com/kubernetes/kubernetes/pull/43699), [@colemickens](https://github.com/colemickens))

- `docker_operations` -> `docker_operations_total`
- `docker_operations_latency_microseconds` -> `docker_operations_latency_seconds`
- `docker_operations_errors` -> `docker_operations_errors_total`
- `docker_operations_timeout` -> `docker_operations_timeout_total`
- `network_plugin_operations_latency_microseconds` -> `network_plugin_operations_latency_seconds`

* Use separate client for leader election in scheduler to avoid starving leader election by regular scheduler operations. ([#53793](https://github.com/kubernetes/kubernetes/pull/53793), [@wojtek-t](https://github.com/wojtek-t))

  The incorrect cost was evident when the result of a function was used in subsequent operations. ([#119800](https://github.com/kubernetes/kubernetes/pull/119800), [@jpbetz](https://github.com/jpbetz)) [SIG API Machinery, Auth and Cloud Provider]

* GCE provider: Log full contents of long operations ([#26962](https://github.com/kubernetes/kubernetes/pull/26962), [@zmerlynn](https://github.com/zmerlynn))

- 'Priority and Fairness has introduced a new feature called _borrowing_ that allows an API priority level
  to borrow a number of seats from other priority level(s). As a cluster operator, you can enable borrowing
  for a certain priority level configuration object via the two newly introduced fields `lendablePercent`, and