关于java Socket setSendBufferSize的一个坑

今天在测试一个问题的时候希望调整java tcp的接收窗口，于是写了一段简单的测试代码:

        SocketChannel socketChannel = SocketChannel.open(new InetSocketAddress("192.168.0.133", 80));

        socketChannel.socket().setReceiveBufferSize(100);
        socketChannel.socket().setSendBufferSize(100);

        socketChannel.write(ByteBuffer.wrap(
                ("GET / HTTP/1.1\r\n" +
                "Host: 192.168.0.133\r\n" +
                "Accept: */*\r\n" +
                "\r\n").getBytes(Charset.forName("UTF-8"))));
        ByteBuffer byteBuffer = ByteBuffer.allocate(100);
        try {
            while (socketChannel.read(byteBuffer) > 0) {
                byteBuffer.flip();
                String content = Charset.forName("UTF-8").decode(byteBuffer).toString();
                System.out.print(content);
                byteBuffer.clear();           
            }
        } finally {
            socketChannel.close();
        }

本来以为通过将发送和接收窗口设置为100会减小wireshark抓包看到的win， 然而：

第一次抓包结果

这下就有点百思不得骑姐了。查看一下代码，发现在open时同时进行了connect，而后才设置的接收和发送buffer，会不会这里有问题？于是稍微调整了代码一下：

        SocketChannel socketChannel = SocketChannel.open();

        socketChannel.socket().setReceiveBufferSize(100);
        socketChannel.socket().setSendBufferSize(100);

        socketChannel.connect(new InetSocketAddress("192.168.0.133", 80));

        socketChannel.write(ByteBuffer.wrap(
                ("GET / HTTP/1.1\r\n" +
                "Host: 192.168.0.133\r\n" +
                "Accept: */*\r\n" +
                "\r\n").getBytes(Charset.forName("UTF-8"))));
        ByteBuffer byteBuffer = ByteBuffer.allocate(100);
        String content;
        try {
            while (socketChannel.read(byteBuffer) > 0) {
                byteBuffer.flip();
                content = Charset.forName("UTF-8").decode(byteBuffer).toString();
                System.out.print(content);
                byteBuffer.clear();
          }           
        } finally {
            socketChannel.close();
        }

果然再次抓包发现结果符合预期了：

调整后抓包结果