a single byte in the string.
</li>

<li>
The iteration order over maps is not specified
and is not guaranteed to be the same from one iteration to the next.
If a map entry that has not yet been reached is removed during iteration,
the corresponding iteration value will not be produced. If a map entry is
created during iteration, that entry may be produced during the iteration or

- Improved HPA performance when using container-specific resource metrics by optimizing container lookup logic to exit early once the target container is found, reducing unnecessary iterations through all containers in a pod. ([#133415](https://github.com/kubernetes/kubernetes/pull/133415), [@AadiDev005](https://github.com/AadiDev005)) [SIG Apps and Autoscaling]

- List operations against the API now return internal server errors instead of partially complete lists when a value cannot be transformed from storage. The updated behavior is consistent with all other operations that require transforming data from storage such as watch and get. ([#69399](https://github.com/kubernetes/kubernetes/pull/69399), [@mikedanese](h...

* 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))

- 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))

* 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))

With the release of Kubernetes 1.18, an alpha version of CSI Proxy for Windows is getting released. CSI proxy enables non-privileged (pre-approved) containers to perform privileged storage operations on Windows. CSI drivers can now be supported in Windows by leveraging CSI proxy.
SIG Storage made a lot of progress in the 1.18 release.
In particular, the following storage features are moving to GA in Kubernetes 1.18:

* Time for attach/detach retry operations has been decreased from 10-12s to 2-6s ([#56974](https://github.com/kubernetes/kubernetes/pull/56974), [@gnufied](https://github.com/gnufied))

### Auth

* Contexts must be named in kubeconfigs. ([#56769](https://github.com/kubernetes/kubernetes/pull/56769), [@dixudx](https://github.com/dixudx))

- For volumes that allow attaches across multiple nodes, attach and detach operations across different nodes are now executed in parallel. ([#89239](https://github.com/kubernetes/kubernetes/pull/89239), [@verult](https://github.com/verult)) [SIG Apps, Node and Storage]