【Zookeeper】源码分析之网络通信（二）之NIOServerCnxn

一、前言

　　前面介绍了ServerCnxn，下面开始学习NIOServerCnxn。

二、NIOServerCnxn源码分析

　　2.1 类的继承关系

public class NIOServerCnxn extends ServerCnxn {}

　　说明：NIOServerCnxn继承了ServerCnxn抽象类，使用NIO来处理与客户端之间的通信，使用单线程处理。

　　2.2 类的内部类

　　1. SendBufferWriter类　

    private class SendBufferWriter extends Writer {
        private StringBuffer sb = new StringBuffer();
        
        /**
         * Check if we are ready to send another chunk.
         * @param force force sending, even if not a full chunk
         */
        // 是否准备好发送另一块
        private void checkFlush(boolean force) {
            if ((force && sb.length() > 0) || sb.length() > 2048) { // 当强制发送并且sb大小大于0，或者sb大小大于2048即发送缓存
                sendBufferSync(ByteBuffer.wrap(sb.toString().getBytes()));
                // clear our internal buffer
                sb.setLength(0);
            }
        }

        @Override
        public void close() throws IOException {
            if (sb == null) return;
            // 关闭之前需要强制性发送缓存
            checkFlush(true);
            sb = null; // clear out the ref to ensure no reuse
        }

        @Override
        public void flush() throws IOException {
            checkFlush(true);
        }

        @Override
        public void write(char[] cbuf, int off, int len) throws IOException {
            sb.append(cbuf, off, len);
            checkFlush(false);
        }
    }

　　说明：该类用来将给客户端的响应进行分块，其核心方法是checkFlush方法，其源码如下

        private void checkFlush(boolean force) {
            if ((force && sb.length() > 0) || sb.length() > 2048) { // 当强制发送并且sb大小大于0，或者sb大小大于2048即发送缓存
                sendBufferSync(ByteBuffer.wrap(sb.toString().getBytes()));
                // clear our internal buffer
                sb.setLength(0);
            }
        }

　　说明：当需要强制发送时，sb缓冲中只要有内容就会同步发送，或者是当sb的大小超过2048（块）时就需要发送，其会调用NIOServerCnxn的sendBufferSync方法，该之后会进行分析，然后再清空sb缓冲。

　　2. CommandThread类　　

    private abstract class CommandThread extends Thread {
        PrintWriter pw;
        
        CommandThread(PrintWriter pw) {
            this.pw = pw;
        }
        
        public void run() {
            try {
                commandRun();
            } catch (IOException ie) {
                LOG.error("Error in running command ", ie);
            } finally {
                cleanupWriterSocket(pw);
            }
        }
        
        public abstract void commandRun() throws IOException;
    }

　　说明：该类用于处理ServerCnxn中的定义的命令，其主要逻辑定义在commandRun方法中，在子类中各自实现，这是一种典型的工厂方法，每个子类对应着一个命令，每个命令使用单独的线程进行处理。

　　2.3 类的属性　　

public class NIOServerCnxn extends ServerCnxn {
    // 日志
    static final Logger LOG = LoggerFactory.getLogger(NIOServerCnxn.class);

    // ServerCnxn工厂
    NIOServerCnxnFactory factory;

    // 针对面向流的连接套接字的可选择通道
    final SocketChannel sock;

    // 表示 SelectableChannel 在 Selector 中注册的标记
    private final SelectionKey sk;

    // 初始化标志
    boolean initialized;

    // 分配四个字节缓冲区
    ByteBuffer lenBuffer = ByteBuffer.allocate(4);

    // 赋值incomingBuffer
    ByteBuffer incomingBuffer = lenBuffer;

    // 缓冲队列
    LinkedBlockingQueue<ByteBuffer> outgoingBuffers = new LinkedBlockingQueue<ByteBuffer>();

    // 会话超时时间
    int sessionTimeout;

    // ZooKeeper服务器
    private final ZooKeeperServer zkServer;

    /**
     * The number of requests that have been submitted but not yet responded to.
     */
    // 已经被提交但还未响应的请求数量
    int outstandingRequests;

    /**
     * This is the id that uniquely identifies the session of a client. Once
     * this session is no longer active, the ephemeral nodes will go away.
     */
    // 会话ID
    long sessionId;

    // 下个会话ID
    static long nextSessionId = 1;
    int outstandingLimit = 1;
    
    private static final String ZK_NOT_SERVING =
        "This ZooKeeper instance is not currently serving requests";
    
    private final static byte fourBytes[] = new byte[4];
}    

　　说明：NIOServerCnxn维护了服务器与客户端之间的Socket通道、用于存储传输内容的缓冲区、会话ID、ZooKeeper服务器等。

　　2.4 类的构造函数　　

    public NIOServerCnxn(ZooKeeperServer zk, SocketChannel sock,
            SelectionKey sk, NIOServerCnxnFactory factory) throws IOException {
        this.zkServer = zk;
        this.sock = sock;
        this.sk = sk;
        this.factory = factory;
        if (this.factory.login != null) {
            this.zooKeeperSaslServer = new ZooKeeperSaslServer(factory.login);
        }
        if (zk != null) { 
            outstandingLimit = zk.getGlobalOutstandingLimit();
        }
        sock.socket().setTcpNoDelay(true);
        /* set socket linger to false, so that socket close does not
         * block */
        // 设置linger为false，以便在socket关闭时不会阻塞
        sock.socket().setSoLinger(false, -1);
        // 获取IP地址
        InetAddress addr = ((InetSocketAddress) sock.socket()
                .getRemoteSocketAddress()).getAddress();
        // 认证信息中添加IP地址
        authInfo.add(new Id("ip", addr.getHostAddress()));
        // 设置感兴趣的操作类型
        sk.interestOps(SelectionKey.OP_READ);
    }

　　说明：在构造函数中会对Socket通道进行相应设置，如设置TCP连接无延迟、获取客户端的IP地址并将此信息进行记录，方便后续认证，最后设置SelectionKey感兴趣的操作类型为READ。

　　2.5 核心函数分析

　　1. sendBuffer函数　　

    public void sendBuffer(ByteBuffer bb) {
        try {
            if (bb != ServerCnxnFactory.closeConn) { // 不关闭连接
                // We check if write interest here because if it is NOT set,
                // nothing is queued, so we can try to send the buffer right
                // away without waking up the selector
                // 首先检查interestOps中是否存在WRITE操作，如果没有
                // 则表示直接发送缓冲而不必先唤醒selector
                if ((sk.interestOps() & SelectionKey.OP_WRITE) == 0) { // 不为write操作
                    try {
                        // 将缓冲写入socket
                        sock.write(bb);
                    } catch (IOException e) {
                        // we are just doing best effort right now
                    }
                }
                // if there is nothing left to send, we are done
                if (bb.remaining() == 0) { // bb中的内容已经被全部读取
                    // 统计发送包信息（调用ServerCnxn方法）
                    packetSent();
                    return;
                }
            }

            synchronized(this.factory){ // 同步块
                // Causes the first selection operation that has not yet returned to return immediately
                // 让第一个还没返回（阻塞）的selection操作马上返回结果
                sk.selector().wakeup();
                if (LOG.isTraceEnabled()) {
                    LOG.trace("Add a buffer to outgoingBuffers, sk " + sk
                            + " is valid: " + sk.isValid());
                }
                // 将缓存添加至队列
                outgoingBuffers.add(bb);
                if (sk.isValid()) { // key是否合法
                    // 将写操作添加至感兴趣的集合
                    sk.interestOps(sk.interestOps() | SelectionKey.OP_WRITE);
                }
            }
            
        } catch(Exception e) {
            LOG.error("Unexpected Exception: ", e);
        }
    }

　　说明：该函数将缓冲写入socket中，其大致处理可以分为两部分，首先会判断ByteBuffer是否为关闭连接的信号，并且当感兴趣的集合中没有写操作时，其会立刻将缓存写入socket，步骤如下

            if (bb != ServerCnxnFactory.closeConn) { // 不关闭连接
                // We check if write interest here because if it is NOT set,
                // nothing is queued, so we can try to send the buffer right
                // away without waking up the selector
                // 首先检查interestOps中是否存在WRITE操作，如果没有
                // 则表示直接发送缓冲而不必先唤醒selector
                if ((sk.interestOps() & SelectionKey.OP_WRITE) == 0) { // 不为write操作
                    try {
                        // 将缓冲写入socket
                        sock.write(bb);
                    } catch (IOException e) {
                        // we are just doing best effort right now
                    }
                }
                // if there is nothing left to send, we are done
                if (bb.remaining() == 0) { // bb中的内容已经被全部读取
                    // 统计发送包信息（调用ServerCnxn方法）
                    packetSent();
                    return;
                }
            }

　　当缓冲区被正常的写入到socket后，会直接返回，然而，当原本就对写操作感兴趣时，其会走如下流程

            synchronized(this.factory){ // 同步块
                // Causes the first selection operation that has not yet returned to return immediately
                // 让第一个还没返回（阻塞）的selection操作马上返回结果
                sk.selector().wakeup();
                if (LOG.isTraceEnabled()) {
                    LOG.trace("Add a buffer to outgoingBuffers, sk " + sk
                            + " is valid: " + sk.isValid());
                }
                // 将缓存添加至队列
                outgoingBuffers.add(bb);
                if (sk.isValid()) { // key是否合法
                    // 将写操作添加至感兴趣的集合
                    sk.interestOps(sk.interestOps() | SelectionKey.OP_WRITE);
                }
            }

　　首先会唤醒上个被阻塞的selection操作，然后将缓冲添加至outgoingBuffers队列中，后续再进行发送。

　　2. doIO函数　

    void doIO(SelectionKey k) throws InterruptedException {
        try {
            if (isSocketOpen() == false) { // socket未开启
                LOG.warn("trying to do i/o on a null socket for session:0x"
                         + Long.toHexString(sessionId));

                return;
            }
            if (k.isReadable()) { // key可读
                // 将内容从socket写入incoming缓冲
                int rc = sock.read(incomingBuffer);
                if (rc < 0) { // 流结束异常，无法从客户端读取数据
                    throw new EndOfStreamException(
                            "Unable to read additional data from client sessionid 0x"
                            + Long.toHexString(sessionId)
                            + ", likely client has closed socket");
                }
                if (incomingBuffer.remaining() == 0) { // 缓冲区已经写满
                    boolean isPayload;
                    // 读取下个请求
                    if (incomingBuffer == lenBuffer) { // start of next request
                        // 翻转缓冲区，可读
                        incomingBuffer.flip();
                        // 读取lenBuffer的前四个字节，当读取的是内容长度时则为true，否则为false
                        isPayload = readLength(k);
                        // 清除缓冲
                        incomingBuffer.clear();
                    } else { // 不等，因为在readLength中根据Len已经重新分配了incomingBuffer
                        // continuation
                        isPayload = true;
                    }
                    if (isPayload) { // 不为四个字母，为实际内容    // not the case for 4letterword
                        // 读取内容
                        readPayload();
                    }
                    else { // 四个字母，为四字母的命令
                        // four letter words take care
                        // need not do anything else
                        return;
                    }
                }
            }
            if (k.isWritable()) { // key可写
                // ZooLog.logTraceMessage(LOG,
                // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK
                // "outgoingBuffers.size() = " +
                // outgoingBuffers.size());
                if (outgoingBuffers.size() > 0) {
                    // ZooLog.logTraceMessage(LOG,
                    // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK,
                    // "sk " + k + " is valid: " +
                    // k.isValid());

                    /*
                     * This is going to reset the buffer position to 0 and the
                     * limit to the size of the buffer, so that we can fill it
                     * with data from the non-direct buffers that we need to
                     * send.
                     */
                    // 分配的直接缓冲
                    ByteBuffer directBuffer = factory.directBuffer;
                    // 清除缓冲
                    directBuffer.clear();

                    for (ByteBuffer b : outgoingBuffers) { // 遍历
                        if (directBuffer.remaining() < b.remaining()) { // directBuffer的剩余空闲长度小于b的剩余空闲长度
                            /*
                             * When we call put later, if the directBuffer is to
                             * small to hold everything, nothing will be copied,
                             * so we've got to slice the buffer if it's too big.
                             */
                            // 缩小缓冲至directBuffer的大小
                            b = (ByteBuffer) b.slice().limit(
                                    directBuffer.remaining());
                        }
                        /*
                         * put() is going to modify the positions of both
                         * buffers, put we don't want to change the position of
                         * the source buffers (we'll do that after the send, if
                         * needed), so we save and reset the position after the
                         * copy
                         */
                        // 记录b的当前position
                        int p = b.position();
                        // 将b写入directBuffer
                        directBuffer.put(b);
                        // 设置回b的原来的position
                        b.position(p);
                        if (directBuffer.remaining() == 0) { // 已经写满
                            break;
                        }
                    }
                    /*
                     * Do the flip: limit becomes position, position gets set to
                     * 0. This sets us up for the write.
                     */
                    // 翻转缓冲区，可读
                    directBuffer.flip();

                    // 将directBuffer的内容写入socket
                    int sent = sock.write(directBuffer);
                    ByteBuffer bb;

                    // Remove the buffers that we have sent
                    while (outgoingBuffers.size() > 0) { // outgoingBuffers中还存在Buffer
                        // 取队首元素，但并不移出
                        bb = outgoingBuffers.peek();
                        if (bb == ServerCnxnFactory.closeConn) { // 关闭连接，抛出异常
                            throw new CloseRequestException("close requested");
                        }
                        
                        // bb还剩余多少元素没有被发送
                        int left = bb.remaining() - sent;
                        if (left > 0) { // 存在元素未被发送
                            /*
                             * We only partially sent this buffer, so we update
                             * the position and exit the loop.
                             */
                            // 更新bb的position
                            bb.position(bb.position() + sent);
                            break;
                        }
                        // 发送包，调用ServerCnxn方法
                        packetSent();
                        /* We've sent the whole buffer, so drop the buffer */
                        // 已经发送完buffer的所有内容，移除buffer
                        sent -= bb.remaining();
                        outgoingBuffers.remove();
                    }
                    // ZooLog.logTraceMessage(LOG,
                    // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK, "after send,
                    // outgoingBuffers.size() = " + outgoingBuffers.size());
                }

                synchronized(this.factory){ // 同步块
                    if (outgoingBuffers.size() == 0) { // outgoingBuffers不存在buffer
                        if (!initialized
                                && (sk.interestOps() & SelectionKey.OP_READ) == 0) { // 未初始化并且无读请求
                            throw new CloseRequestException("responded to info probe");
                        }
                        // 重置感兴趣的集合
                        sk.interestOps(sk.interestOps()
                                & (~SelectionKey.OP_WRITE));
                    } else { // 重置感兴趣的集合
                        sk.interestOps(sk.interestOps()
                                | SelectionKey.OP_WRITE);
                    }
                }
            }
        } catch (CancelledKeyException e) {
            LOG.warn("Exception causing close of session 0x"
                    + Long.toHexString(sessionId)
                    + " due to " + e);
            if (LOG.isDebugEnabled()) {
                LOG.debug("CancelledKeyException stack trace", e);
            }
            close();
        } catch (CloseRequestException e) {
            // expecting close to log session closure
            close();
        } catch (EndOfStreamException e) {
            LOG.warn("caught end of stream exception",e); // tell user why

            // expecting close to log session closure
            close();
        } catch (IOException e) {
            LOG.warn("Exception causing close of session 0x"
                    + Long.toHexString(sessionId)
                    + " due to " + e);
            if (LOG.isDebugEnabled()) {
                LOG.debug("IOException stack trace", e);
            }
            close();
        }
    }

　　说明：该函数主要是进行IO处理，当传入的SelectionKey是可读时，其处理流程如下

            if (k.isReadable()) { // key可读
                // 将内容从socket写入incoming缓冲
                int rc = sock.read(incomingBuffer);
                if (rc < 0) { // 流结束异常，无法从客户端读取数据
                    throw new EndOfStreamException(
                            "Unable to read additional data from client sessionid 0x"
                            + Long.toHexString(sessionId)
                            + ", likely client has closed socket");
                }
                if (incomingBuffer.remaining() == 0) { // 缓冲区已经写满
                    boolean isPayload;
                    // 读取下个请求
                    if (incomingBuffer == lenBuffer) { // start of next request
                        // 翻转缓冲区，可读
                        incomingBuffer.flip();
                        // 读取lenBuffer的前四个字节，当读取的是内容长度时则为true，否则为false
                        isPayload = readLength(k);
                        // 清除缓冲
                        incomingBuffer.clear();
                    } else { // 不等，因为在readLength中根据Len已经重新分配了incomingBuffer
                        // continuation
                        isPayload = true;
                    }
                    if (isPayload) { // 不为四个字母，为实际内容    // not the case for 4letterword
                        // 读取内容
                        readPayload();
                    }
                    else { // 四个字母，为四字母的命令
                        // four letter words take care
                        // need not do anything else
                        return;
                    }
                }
            }

　　说明：首先从socket中将数据读入incomingBuffer中，再判断incomingBuffer是否与lenBuffer相等，若相等，则表示读取的是一个四个字母的命令，否则表示读取的是具体内容的长度，因为在readLength函数会根据socket中内容的长度重新分配incomingBuffer。其中，readLength函数的源码如下　

    private boolean readLength(SelectionKey k) throws IOException {
        // Read the length, now get the buffer
        // 读取position之后的四个字节
        int len = lenBuffer.getInt();
        if (!initialized && checkFourLetterWord(sk, len)) { // 未初始化并且是四个字母组成的命令
            return false;
        }
        if (len < 0 || len > BinaryInputArchive.maxBuffer) {
            throw new IOException("Len error " + len);
        }
        if (zkServer == null) {
            throw new IOException("ZooKeeperServer not running");
        }
        // 重新分配len长度的缓冲
        incomingBuffer = ByteBuffer.allocate(len);
        return true;
    }

　　说明：首先会读取lenBuffer缓冲的position之后的四个字节，然后判断其是否是四字母的命令或者是长整形（具体内容的长度），之后再根据长度重新分配incomingBuffer大小。

　　同时，在调用完readLength后，会知道是否为内容，若为内容，则会调用readPayload函数来读取内容，其源码如下　　

    private void readPayload() throws IOException, InterruptedException {
        // 表示还未读取完socket中内容
        if (incomingBuffer.remaining() != 0) { // have we read length bytes?
            // 将socket的内容读入缓冲
            int rc = sock.read(incomingBuffer); // sock is non-blocking, so ok
            if (rc < 0) { // 流结束异常，无法从客户端读取数据
                throw new EndOfStreamException(
                        "Unable to read additional data from client sessionid 0x"
                        + Long.toHexString(sessionId)
                        + ", likely client has closed socket");
            }
        }
        
        // 表示已经读取完了Socket中内容
        if (incomingBuffer.remaining() == 0) { // have we read length bytes?
            // 接收到packet
            packetReceived();
            // 翻转缓冲区
            incomingBuffer.flip();
            if (!initialized) { // 未初始化
                // 读取连接请求
                readConnectRequest();
            } else {
                // 读取请求
                readRequest();
            }
            // 清除缓冲
            lenBuffer.clear();
            // 赋值incomingBuffer，即清除incoming缓冲
            incomingBuffer = lenBuffer;
        }
    }

　　说明：首先会将socket中的实际内容写入incomingBuffer中（已经重新分配大小），当读取完成后，则更新接收的包统计信息，之后再根据是否初始化了还确定读取连接请求还是直接请求，最后会清除缓存，并重新让incomingBuffer与lenBuffer相等，表示该读取过程结束。

　　而当doIO中的key为可写时，其处理流程如下　

            if (k.isWritable()) { // key可写
                // ZooLog.logTraceMessage(LOG,
                // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK
                // "outgoingBuffers.size() = " +
                // outgoingBuffers.size());
                if (outgoingBuffers.size() > 0) {
                    // ZooLog.logTraceMessage(LOG,
                    // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK,
                    // "sk " + k + " is valid: " +
                    // k.isValid());

                    /*
                     * This is going to reset the buffer position to 0 and the
                     * limit to the size of the buffer, so that we can fill it
                     * with data from the non-direct buffers that we need to
                     * send.
                     */
                    // 分配的直接缓冲
                    ByteBuffer directBuffer = factory.directBuffer;
                    // 清除缓冲
                    directBuffer.clear();

                    for (ByteBuffer b : outgoingBuffers) { // 遍历
                        if (directBuffer.remaining() < b.remaining()) { // directBuffer的剩余空闲长度小于b的剩余空闲长度
                            /*
                             * When we call put later, if the directBuffer is to
                             * small to hold everything, nothing will be copied,
                             * so we've got to slice the buffer if it's too big.
                             */
                            // 缩小缓冲至directBuffer的大小
                            b = (ByteBuffer) b.slice().limit(
                                    directBuffer.remaining());
                        }
                        /*
                         * put() is going to modify the positions of both
                         * buffers, put we don't want to change the position of
                         * the source buffers (we'll do that after the send, if
                         * needed), so we save and reset the position after the
                         * copy
                         */
                        // 记录b的当前position
                        int p = b.position();
                        // 将b写入directBuffer
                        directBuffer.put(b);
                        // 设置回b的原来的position
                        b.position(p);
                        if (directBuffer.remaining() == 0) { // 已经写满
                            break;
                        }
                    }
                    /*
                     * Do the flip: limit becomes position, position gets set to
                     * 0. This sets us up for the write.
                     */
                    // 翻转缓冲区，可读
                    directBuffer.flip();

                    // 将directBuffer的内容写入socket
                    int sent = sock.write(directBuffer);
                    ByteBuffer bb;

                    // Remove the buffers that we have sent
                    while (outgoingBuffers.size() > 0) { // outgoingBuffers中还存在Buffer
                        // 取队首元素，但并不移出
                        bb = outgoingBuffers.peek();
                        if (bb == ServerCnxnFactory.closeConn) { // 关闭连接，抛出异常
                            throw new CloseRequestException("close requested");
                        }
                        
                        // bb还剩余多少元素没有被发送
                        int left = bb.remaining() - sent;
                        if (left > 0) { // 存在元素未被发送
                            /*
                             * We only partially sent this buffer, so we update
                             * the position and exit the loop.
                             */
                            // 更新bb的position
                            bb.position(bb.position() + sent);
                            break;
                        }
                        // 发送包，调用ServerCnxn方法
                        packetSent();
                        /* We've sent the whole buffer, so drop the buffer */
                        // 已经发送完buffer的所有内容，移除buffer
                        sent -= bb.remaining();
                        outgoingBuffers.remove();
                    }
                    // ZooLog.logTraceMessage(LOG,
                    // ZooLog.CLIENT_DATA_PACKET_TRACE_MASK, "after send,
                    // outgoingBuffers.size() = " + outgoingBuffers.size());
                }

　　说明：其首先会判断outgoingBuffers中是否还有Buffer未发送，然后遍历Buffer，为提供IO效率，借助了directBuffer（64K大小），之后每次以directBuffer的大小(64K)来将缓冲的内容写入socket中发送，直至全部发送完成。

三、总结

　　本篇讲解了NIOServerCnxn的处理细节，其主要依托于Java的NIO相关接口来完成IO操作，也谢谢各位园友的观看~