// sequent of operations:
	// 1) pipe.Write()
	// 2) pipe.Close()
	// Now pipe.Close() can return before the data is read on the other end of the pipe and written to the disk
	// Hence an immediate Read() on the file can return incorrect data.
	if b.canClose != nil {
		b.canClose.Wait()
	}

	// Recycle the buffer.
	if b.byteBuf != nil {

So we have a dedicated thread for every socket that just reads frames and dispatches them.

The reader thread must never run application-layer code. Otherwise one slow stream can hold up the entire connection.

    // merged into POM being built. Once the POM builder has this plugin information, versions can be assigned
    // by the POM builder because they will have to be defined in plugin management. Once this is setComplete then it
    // can be passed back so that the default configuration information can be populated.
    //
    // We need to know the specific version so that we can look up the right version of the plugin descriptor

   * The maximum number of permits that can be saved (when the RateLimiter is unused) is defined in
   * terms of time, in this sense: if a RateLimiter is 2qps, and this time is specified as 10
   * seconds, we can save up to 2 * 10 = 20 permits.
   */
  static final class SmoothBursty extends SmoothRateLimiter {
    /** The work (permits) of how many seconds can be saved up if this RateLimiter is unused? */

message RollingUpdateDeployment {
  // The maximum number of pods that can be unavailable during the update.
  // Value can be an absolute number (ex: 5) or a percentage of desired pods (ex: 10%).
  // Absolute number is calculated from percentage by rounding down.
  // This can not be 0 if MaxSurge is 0.
  // By default, a fixed value of 1 is used.
  // Example: when this is set to 30%, the old RC can be scaled down to 70% of desired pods

  ) {
  }

  /**
   * Invoked just prior to initiating a socket connection.
   *
   * This method will be invoked if no existing connection in the [ConnectionPool] can be reused.
   *
   * This can be invoked more than 1 time for a single [Call]. For example, if the response to the
   * [Call.request] is a redirect to a different address, or a connection is retried.
   */
  open fun connectStart(

   *
   * <ul>
   *   <li>You want to assign the same fraction of inputs to each bucket.
   *   <li>When you reduce the number of buckets, you can accept that the most recently added
   *       buckets will be removed first. More concretely, if you are dividing traffic among tasks,
   *       you can decrease the number of tasks from 15 and 10, killing off the final 5 tasks, and

    once you get into the running container so `bazel` can find the `tensorflow`
    workspace).

    you can do this by using the following command. As an example-

    ```bash
    docker run -it --rm -v $PWD:/tmp -w /tmp tensorflow/build:2.15-python3.10
    ```

    Once you have the packages installed, you can run a specific unit test in
    bazel by doing as follows:

    ```bash

This only applies for selector-based policies; namespaced and global policies can be handled without needing to list them out in the Workload API.

## Redirection

As ztunnel aims to transparently encrypt and route users traffic, we need a mechanism to capture all traffic entering and leaving "mesh" pods.
This is a security critical task: if the ztunnel can be bypassed, authorization policies can be bypassed.

Redirection must meet these requirements:

message RollingUpdateDeployment {
  // The maximum number of pods that can be unavailable during the update.
  // Value can be an absolute number (ex: 5) or a percentage of desired pods (ex: 10%).
  // Absolute number is calculated from percentage by rounding down.
  // This can not be 0 if MaxSurge is 0.
  // Defaults to 25%.
  // Example: when this is set to 30%, the old ReplicaSet can be scaled down to 70% of desired pods