* asynchronous operations. You can chain them together manually with calls to methods like {@link
 * Futures#transform(ListenableFuture, Function, Executor) Futures.transform}, but you will often
 * find it easier to use a framework. Frameworks automate the process, often adding features like
 * monitoring, debugging, and cancellation. Examples of frameworks include:
 *
 * <ul>

- Secure master -> node communication ([#11816](https://github.com/kubernetes/kubernetes/issues/11816))
- upgrading master doesn't upgrade kubectl ([#32538](https://github.com/kubernetes/kubernetes/issues/32538))
- Specific error message on failed rolling update issued by older kubectl against 1.4 master ([#32751](https://github.com/kubernetes/kubernetes/issues/32751))

    [`tf.debugging`](https://www.tensorflow.org/versions/master/api_docs/python/tf/debugging),
    [`tf.dtypes`](https://www.tensorflow.org/versions/master/api_docs/python/tf/dtypes),
    [`tf.image`](https://www.tensorflow.org/versions/master/api_docs/python/tf/image),
    [`tf.io`](https://www.tensorflow.org/versions/master/api_docs/python/tf/io),
    [`tf.linalg`](https://www.tensorflow.org/versions/master/api_docs/python/tf/linalg),

  * Use optimized image based on Jessie by default
  * Add support for Ubuntu Wily
  * Master is configured with automatic restart-on-failure, via CloudWatch
  * Bootstrap reworked to be more similar to GCE; better supports reboots/restarts
  * Use an elastic IP for the master by default
  * Experimental support for node spot instances (set NODE\_SPOT\_PRICE=0.05)

#### GCE

### SIG Azure

* The Azure cloud provider now supports standard SKU load balancer and public IP.
`excludeMasterFromStandardLB` defaults to true, which means master nodes are excluded from the standard load balancer. Also note that because all nodes (except master) are added as loadbalancer backends, the standard load balancer doesn't work with the `service.beta.kubernetes.io/azure-load-balancer-mode` annotation.

 * asynchronous operations. You can chain them together manually with calls to methods like {@link
 * Futures#transform(ListenableFuture, Function, Executor) Futures.transform}, but you will often
 * find it easier to use a framework. Frameworks automate the process, often adding features like
 * monitoring, debugging, and cancellation. Examples of frameworks include:
 *
 * <ul>

	// An untyped nil does not need a pointer check, and when
	// _cgoCheckPointer returns the untyped nil the type assertion we
	// are going to insert will fail.  Easier to just skip nil arguments.
	// TODO: Note that this fails if nil is shadowed.
	if id, ok := arg.(*ast.Ident); ok && id.Name == "nil" {
		return false
	}

	return p.hasPointer(f, t, true)
}

     * changing any bin. The operations must not take any action that could even momentarily
     * cause a concurrent read operation to see inconsistent data. This is made easier by the
     * nature of the read operations in Map. For example, no operation can reveal that the table
     * has grown but the threshold has not yet been updated, so there are no atomicity requirements

     * changing any bin. The operations must not take any action that could even momentarily
     * cause a concurrent read operation to see inconsistent data. This is made easier by the
     * nature of the read operations in Map. For example, no operation can reveal that the table
     * has grown but the threshold has not yet been updated, so there are no atomicity requirements

* Bump up glbc version to 0.6.2 ([#25446](https://github.com/kubernetes/kubernetes/pull/25446), [@bprashanth](https://github.com/bprashanth))
* Add "kubectl set image" for easier updating container images (for pods or resources with pod templates).  ([#25509](https://github.com/kubernetes/kubernetes/pull/25509), [@janetkuo](https://github.com/janetkuo))