JAVA RPC (九) netty服务端解析

源码地址：https://gitee.com/a1234567891/koalas-rpc

企业生产级百亿日PV高可用可拓展的RPC框架。理论上并发数量接近服务器带宽，客户端采用thrift协议，服务端支持netty和thrift的TThreadedSelectorServer半同步半异步线程模型，支持动态扩容，服务上下线，权重动态，可用性配置，页面流量统计，支持trace跟踪等，天然接入cat支持数据大盘展示等，持续为个人以及中小型公司提供可靠的RPC框架技术方案，贴上

  @Override
    public void run() {
        try {
            if (Epoll.isAvailable ()) {
                bossGroup = new EpollEventLoopGroup (serverPublisher.bossThreadCount==0?AbstractKoalsServerPublisher.DEFAULT_EVENT_LOOP_THREADS:serverPublisher.bossThreadCount);
                workerGroup = new EpollEventLoopGroup ( serverPublisher.workThreadCount==0? AbstractKoalsServerPublisher.DEFAULT_EVENT_LOOP_THREADS*2:serverPublisher.workThreadCount);
            } else {
                bossGroup = new NioEventLoopGroup (serverPublisher.bossThreadCount==0?AbstractKoalsServerPublisher.DEFAULT_EVENT_LOOP_THREADS:serverPublisher.bossThreadCount);
                workerGroup = new NioEventLoopGroup ( serverPublisher.workThreadCount==0? AbstractKoalsServerPublisher.DEFAULT_EVENT_LOOP_THREADS*2:serverPublisher.workThreadCount );
            }
            executorService = KoalasThreadedSelectorWorkerExcutorUtil.getWorkerExecutorWithQueue (serverPublisher.koalasThreadCount==0?AbstractKoalsServerPublisher.DEFAULT_KOALAS_THREADS:serverPublisher.koalasThreadCount,serverPublisher.koalasThreadCount==0?AbstractKoalsServerPublisher.DEFAULT_KOALAS_THREADS:serverPublisher.koalasThreadCount,serverPublisher.workQueue,new KoalasDefaultThreadFactory (serverPublisher.serviceInterface.getName ()));

            ServerBootstrap b = new ServerBootstrap ();
            b.group ( bossGroup, workerGroup ).channel ( workerGroup instanceof EpollEventLoopGroup ? EpollServerSocketChannel.class : NioServerSocketChannel.class )
                    .handler ( new LoggingHandler ( LogLevel.INFO ) )
                    .childHandler ( new NettyServerInitiator (serverPublisher,executorService))
                    .option ( ChannelOption.SO_BACKLOG, 1024 )
                    .option ( ChannelOption.SO_REUSEADDR, true )
                    .option ( ChannelOption.SO_KEEPALIVE, true );
            Channel ch = b.bind ( serverPublisher.port ).sync ().channel ();
            Runtime.getRuntime().addShutdownHook(new Thread(){
                @Override
                public void run(){
                    logger.info ( "Shutdown by Runtime" );
                    if(zookeeperServer != null){
                        zookeeperServer.destroy ();
                    }
                    logger.info ( "wait for service over 3000ms" );
                    try {
                        Thread.sleep ( 3000 );
                    } catch (Exception e) {
                    }
                    if(executorService!=null){
                        executorService.shutdown ();
                    }
                    if(bossGroup != null) bossGroup.shutdownGracefully ();
                    if(workerGroup != null) workerGroup.shutdownGracefully ();
                }
            });

            if(StringUtils.isNotEmpty ( serverPublisher.zkpath )){
                ZookServerConfig zookServerConfig = new ZookServerConfig ( serverPublisher.zkpath,serverPublisher.serviceInterface.getName (),serverPublisher.env,serverPublisher.port,serverPublisher.weight,"netty" );
                zookeeperServer = new ZookeeperServer ( zookServerConfig );
                zookeeperServer.init ();
            }
        } catch ( Exception e){
            logger.error ( "NettyServer start faid !",e );
            if(bossGroup != null) bossGroup.shutdownGracefully ();
            if(workerGroup != null) workerGroup.shutdownGracefully ();
        }

        logger.info("netty server init success server={}",serverPublisher);

    }

首先开启NIO服务，由系统内核来判断是否支持epoll-EpollEventLoopGroup,如果不支持epoll采用IO多路复用的方式EpollEventLoopGroup，然后声明一个用户自定义线程池，这里有不清楚的读者肯定会问，netty本身支持连接线程和IO线程，为什么还要自定义声明自定义线程池，原因是假设在IO线程池中做的业务非常复杂，大量耗时，这样就会阻塞了netty线程的IO处理速度，影响吞吐量，这也就是reactor模型的设计理念，不让业务干扰连接线程和IO读写线程。NettyServerInitiator就是实际处理的业务handle了。

 Runtime.getRuntime().addShutdownHook(new Thread(){
                @Override
                public void run(){
                    logger.info ( "Shutdown by Runtime" );
                    if(zookeeperServer != null){
                        zookeeperServer.destroy ();
                    }
                    logger.info ( "wait for service over 3000ms" );
                    try {
                        Thread.sleep ( 3000 );
                    } catch (Exception e) {
                    }
                    if(executorService!=null){
                        executorService.shutdown ();
                    }
                    if(bossGroup != null) bossGroup.shutdownGracefully ();
                    if(workerGroup != null) workerGroup.shutdownGracefully ();
                }
            });

手动关闭钩子，服务关闭的时候要主动关闭节点信息。下面来看一下hander拦截器

package netty.initializer;

import io.netty.channel.ChannelInitializer;
import io.netty.channel.socket.SocketChannel;
import netty.hanlder.KoalasDecoder;
import netty.hanlder.KoalasEncoder;
import netty.hanlder.KoalasHandler;
import org.apache.thrift.TProcessor;
import server.config.AbstractKoalsServerPublisher;

import java.util.concurrent.ExecutorService;
/**
 * Copyright (C) 2018
 * All rights reserved
 * User: yulong.zhang
 * Date:2018年11月23日11:13:33
 */
public class NettyServerInitiator extends ChannelInitializer<SocketChannel> {

    private ExecutorService executorService;

    private  AbstractKoalsServerPublisher serverPublisher;

    public NettyServerInitiator(AbstractKoalsServerPublisher serverPublisher,ExecutorService executorService){
        this.serverPublisher = serverPublisher;
        this.executorService = executorService;
    }

    @Override
    protected void initChannel(SocketChannel ch) {
        ch.pipeline ().addLast ( "decoder",new KoalasDecoder () );
        ch.pipeline ().addLast ( "encoder",new KoalasEncoder ());
        ch.pipeline ().addLast ( "handler",new KoalasHandler (serverPublisher,executorService) );
    }

}

decode负责拆包。encoder负责装包，handler是真正业务处理的逻辑，所有的业务处理都在这里的线程池中运行，结果通过ChannelHandlerContext 异步的返回给client端，通过这种方式真正的实现了reactor
下面我们看看拆包处理

package netty.hanlder;

import io.netty.buffer.ByteBuf;
import io.netty.channel.ChannelHandlerContext;
import io.netty.handler.codec.ByteToMessageDecoder;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import server.KoalasServerPublisher;

import java.util.List;
/**
 * Copyright (C) 2018
 * All rights reserved
 * User: yulong.zhang
 * Date:2018年11月23日11:13:33
 */
public class KoalasDecoder extends ByteToMessageDecoder {

    private final static Logger logger = LoggerFactory.getLogger ( KoalasDecoder.class );

    @Override
    protected void decode(ChannelHandlerContext ctx, ByteBuf in, List<Object> out) {

        try {
            if (in.readableBytes () < 4) {
                return;
            }

            in.markReaderIndex ();
            byte[] b = new byte[4];
            in.readBytes ( b );

            int length = decodeFrameSize ( b );

            if (in.readableBytes () < length) {
                //reset the readerIndex
                in.resetReaderIndex ();
                return;
            }

            in.resetReaderIndex ();
            ByteBuf fream = in.readRetainedSlice ( 4 + length );
            in.resetReaderIndex ();

            in.skipBytes ( 4 + length );
            out.add ( fream );
        } catch (Exception e) {
            logger.error ( "decode error",e );
        }

    }

    public static final int decodeFrameSize(byte[] buf) {
        return (buf[0] & 255) << 24 | (buf[1] & 255) << 16 | (buf[2] & 255) << 8 | buf[3] & 255;
    }
}

通过读取四个字节的长度来决定消息体长度，然后根据消息体长度来读取所有的字节流数据。decodeFrameSize方法将四个字节流转成int类型。KoalasHandler处理器逻辑比较复杂，我们只看核心的内容，首先通过thrift解析字节流来获取transport

            ByteArrayInputStream inputStream = new ByteArrayInputStream ( b );
            ByteArrayOutputStream outputStream = new ByteArrayOutputStream (  );

            TIOStreamTransport tioStreamTransportInput = new TIOStreamTransport (  inputStream);
            TIOStreamTransport tioStreamTransportOutput = new TIOStreamTransport (  outputStream);

            TKoalasFramedTransport inTransport = new TKoalasFramedTransport ( tioStreamTransportInput,2048000 );
            inTransport.setReadMaxLength_ ( maxLength );
            TKoalasFramedTransport outTransport = new TKoalasFramedTransport ( tioStreamTransportOutput,2048000,ifUserProtocol );

最终扔到线程池里去执行，将当前IO线程释放给下一个任务。

           try {
                executorService.execute ( new NettyRunable (  ctx,in,out,outputStream,localTprocessor,b,privateKey,publicKey,className,methodName,koalasTrace,cat));
            } catch (RejectedExecutionException e){
                logger.error ( e.getMessage ()+ErrorType.THREAD+",className:" +className,e );
                handlerException(b,ctx,e,ErrorType.THREAD,privateKey,publicKey,thriftNative);
            }

RejectedExecutionException来负责当线程池不够用的时候返回给client端异常，因为server端的业务处理能力有限，所以这里适当的做了一下服务端保护防止雪崩的问题。当发现server端有大量的

RejectedExecutionException抛出，说明单机已经无法满足业务请求了，需要横向拓展机器来进行负载均衡。用户实际的业务执行是在Runable里，我们看看他到底做了什么

           try {
                tprocessor.process ( in,out );
                ctx.writeAndFlush (outputStream);
                if(transaction!=null && cat)
                    transaction.setStatus ( Transaction.SUCCESS );
            } catch (Exception e) {
                if(transaction!=null && cat)
                    transaction.setStatus ( e );
                logger.error ( e.getMessage () + ErrorType.APPLICATION+",className:"+className,e );
                handlerException(this.b,ctx,e,ErrorType.APPLICATION,privateKey,publicKey,thriftNative);
            }

通过thrift的process来执行业务逻辑，将结果通过ctx.writeAndFlush (outputStream)，返回给client端。在catch里处理当出现异常之后返回给client端异常结果。这样netty server的实现就全部结束了，thrift服务端解析相关内容我们下一篇来说，里面当中有很多细节需要读者是跟着源码阅读，如果有问题欢迎加群825199617来交流，更多spring，spring mvc，aop，jdk等源码交流等你来！