Usage:  "shutdown timeout to gracefully shutdown server (DEPRECATED)",
		EnvVar: "MINIO_SHUTDOWN_TIMEOUT",
		Hidden: true,
	},

	cli.DurationFlag{
		Name:   "idle-timeout",
		Value:  xhttp.DefaultIdleTimeout,
		Usage:  "idle timeout is the maximum amount of time to wait for the next request when keep-alive are enabled",
		EnvVar: "MINIO_IDLE_TIMEOUT",
		Hidden: true,
	},
	cli.DurationFlag{

pkg net, type KeepAliveConfig struct #62254
pkg net, type KeepAliveConfig struct, Count int #62254
pkg net, type KeepAliveConfig struct, Enable bool #62254
pkg net, type KeepAliveConfig struct, Idle time.Duration #62254
pkg net, type KeepAliveConfig struct, Interval time.Duration #62254
pkg net, type ListenConfig struct, KeepAliveConfig KeepAliveConfig #62254
pkg net/http, func ParseCookie(string) ([]*Cookie, error) #66008

Blocking APIs may be inefficient because you hold a thread idle while waiting on the network. Threads are expensive because they have both a memory overhead and a context-switching overhead.

#### Framed protocols

        responseBodyComplete()
      }
      closed = true
    }
  }

  companion object {
    private const val NO_CHUNK_YET = -1L

    private const val STATE_IDLE = 0 // Idle connections are ready to write request headers.
    private const val STATE_OPEN_REQUEST_BODY = 1
    private const val STATE_WRITING_REQUEST_BODY = 2
    private const val STATE_READ_RESPONSE_HEADERS = 3

This can help ensure that clients do not hold onto pooled connections for too long.
This is not strictly required; Ztunnel clients have both keepalives (which will start to fail once the other end is shut down) and pool idle timeouts (which will close the connection when it has had no activity for some period; because the backend is gone, there will be no activity).
However, it would be ideal to give immediate feedback rather than relying on these mechanisms.

   * time which we translate to storedPermits. (We increase storedPermits with the amount of permits
   * that would have been produced in that idle time). So, if rate == 1 permit per second, and
   * arrivals come exactly one second after the previous, then storedPermits is _never_ increased --
   * we would only increase it for arrivals _later_ than the expected one second.

      encodings[ 0x13] = encoding // Device Control 3 (oft. XOFF)
      encodings[ 0x14] = encoding // Device Control 4
      encodings[ 0x15] = encoding // Negative Acknowledgment
      encodings[ 0x16] = encoding // Synchronous idle
      encodings[ 0x17] = encoding // End of Transmission Block
      encodings[ 0x18] = encoding // Cancel
      encodings[ 0x19] = encoding // End of Medium
      encodings[ 0x1a] = encoding // Substitute

	return c.CallWithHTTPMethod(ctx, http.MethodPost, rpcMethod, values, body, length)
}

// Close closes all idle connections of the underlying http client
func (c *Client) Close() {
	atomic.StoreInt32(&c.connected, closed)
}

// NewClient - returns new REST client.

   * response to the sender.
   *
   * @since 8.0
   */
  public static final byte NAK = 21;

  /**
   * Synchronous Idle: A communication control character used by a synchronous transmission system
   * in the absence of any other character to provide a signal from which synchronism may be
   * achieved or retained.
   *
   * @since 8.0

 * your HTTP calls. This is because each client holds its own connection pool and thread pools.
 * Reusing connections and threads reduces latency and saves memory. Conversely, creating a client
 * for each request wastes resources on idle pools.
 *
 * Use `new OkHttpClient()` to create a shared instance with the default settings:
 *
 * ```java
 * // The singleton HTTP client.
 * public final OkHttpClient client = new OkHttpClient();
 * ```
 *