dubbo源码解析(二十四)远程调用——dubbo协议

5次阅读

共计 52232 个字符,预计需要花费 131 分钟才能阅读完成。

远程调用——dubbo 协议
目标:介绍远程调用中跟 dubbo 协议相关的设计和实现,介绍 dubbo-rpc-dubbo 的源码。
前言
Dubbo 缺省协议采用单一长连接和 NIO 异步通讯,适合于小数据量大并发的服务调用,以及服务消费者机器数远大于服务提供者机器数的情况。反之,Dubbo 缺省协议不适合传送大数据量的服务,比如传文件,传视频等,除非请求量很低。这是官方文档的原话,并且官方文档还介绍了为什么使用单一长连接和 NIO 异步通讯以及为什么不适合传输大数据的服务。我就不赘述了。
我们先来看看 dubbo-rpc-dubbo 下的包结构:

filter:该包下面是对于 dubbo 协议独有的两个过滤器
status:该包下是做了对于服务和线程池状态的检测
telnet:该包下是对于 telnet 命令的支持
最外层:最外层是 dubbo 协议的核心

源码分析
(一)DubboInvoker
该类是 dubbo 协议独自实现的的 invoker,其中实现了调用方法的三种模式,分别是异步发送、单向发送和同步发送,具体在下面介绍。
1. 属性
/**
* 信息交换客户端数组
*/
private final ExchangeClient[] clients;

/**
* 客户端数组位置
*/
private final AtomicPositiveInteger index = new AtomicPositiveInteger();

/**
* 版本号
*/
private final String version;

/**
* 销毁锁
*/
private final ReentrantLock destroyLock = new ReentrantLock();

/**
* Invoker 对象集合
*/
private final Set<Invoker<?>> invokers;
2.doInvoke
@Override
protected Result doInvoke(final Invocation invocation) throws Throwable {
// rpc 会话域
RpcInvocation inv = (RpcInvocation) invocation;
// 获得方法名
final String methodName = RpcUtils.getMethodName(invocation);
// 把 path 放入到附加值中
inv.setAttachment(Constants.PATH_KEY, getUrl().getPath());
// 把版本号放入到附加值
inv.setAttachment(Constants.VERSION_KEY, version);

// 当前的客户端
ExchangeClient currentClient;
// 如果数组内就一个客户端,则直接取出
if (clients.length == 1) {
currentClient = clients[0];
} else {
// 取模轮询 从数组中取,当取到最后一个时,从头开始
currentClient = clients[index.getAndIncrement() % clients.length];
}
try {
// 是否启用异步
boolean isAsync = RpcUtils.isAsync(getUrl(), invocation);
// 是否是单向发送
boolean isOneway = RpcUtils.isOneway(getUrl(), invocation);
// 获得超时时间
int timeout = getUrl().getMethodParameter(methodName, Constants.TIMEOUT_KEY, Constants.DEFAULT_TIMEOUT);
// 如果是单项发送
if (isOneway) {
boolean isSent = getUrl().getMethodParameter(methodName, Constants.SENT_KEY, false);
// 单向发送只负责发送消息,不等待服务端应答,所以没有返回值
currentClient.send(inv, isSent);
RpcContext.getContext().setFuture(null);
return new RpcResult();
} else if (isAsync) {
// 异步调用
ResponseFuture future = currentClient.request(inv, timeout);
// 保存 future,方便后期处理
RpcContext.getContext().setFuture(new FutureAdapter<Object>(future));
return new RpcResult();
} else {
// 同步调用,等待返回结果
RpcContext.getContext().setFuture(null);
return (Result) currentClient.request(inv, timeout).get();
}
} catch (TimeoutException e) {
throw new RpcException(RpcException.TIMEOUT_EXCEPTION, “Invoke remote method timeout. method: ” + invocation.getMethodName() + “, provider: ” + getUrl() + “, cause: ” + e.getMessage(), e);
} catch (RemotingException e) {
throw new RpcException(RpcException.NETWORK_EXCEPTION, “Failed to invoke remote method: ” + invocation.getMethodName() + “, provider: ” + getUrl() + “, cause: ” + e.getMessage(), e);
}
}
在调用 invoker 的时候,通过远程通信将 Invocation 信息传递给服务端,服务端在接收到该 invocation 信息后,要找到对应的本地方法,然后通过反射执行该方法,将方法的执行结果返回给客户端,在这里,客户端发送有三种模式:

异步发送,也就是当我发送调用后,我不阻塞等待结果,直接返回,将返回的 future 保存到上下文,方便后期使用。
单向发送,执行方法不需要返回结果。
同步发送,执行方法后,等待结果返回,否则一直阻塞。

3.isAvailable
@Override
public boolean isAvailable() {
if (!super.isAvailable())
return false;
for (ExchangeClient client : clients) {
// 只要有一个客户端连接并且不是只读,则表示存活
if (client.isConnected() && !client.hasAttribute(Constants.CHANNEL_ATTRIBUTE_READONLY_KEY)) {
//cannot write == not Available ?
return true;
}
}
return false;
}
该方法是检查服务端是否存活。
4.destroy
@Override
public void destroy() {
// in order to avoid closing a client multiple times, a counter is used in case of connection per jvm, every
// time when client.close() is called, counter counts down once, and when counter reaches zero, client will be
// closed.
if (super.isDestroyed()) {
return;
} else {
// double check to avoid dup close
// 获得销毁锁
destroyLock.lock();
try {
if (super.isDestroyed()) {
return;
}
// 销毁
super.destroy();
// 从集合中移除
if (invokers != null) {
invokers.remove(this);
}
for (ExchangeClient client : clients) {
try {
// 关闭每一个客户端
client.close(ConfigUtils.getServerShutdownTimeout());
} catch (Throwable t) {
logger.warn(t.getMessage(), t);
}
}

} finally {
// 释放锁
destroyLock.unlock();
}
}
}
该方法是销毁服务端,关闭所有连接到远程通信客户端。
(二)DubboExporter
该类继承了 AbstractExporter,是 dubbo 协议中独有的服务暴露者。
/**
* 服务 key
*/
private final String key;

/**
* 服务暴露者集合
*/
private final Map<String, Exporter<?>> exporterMap;

public DubboExporter(Invoker<T> invoker, String key, Map<String, Exporter<?>> exporterMap) {
super(invoker);
this.key = key;
this.exporterMap = exporterMap;
}

@Override
public void unexport() {
super.unexport();
// 从集合中移除该 key
exporterMap.remove(key);
}
其中对于服务暴露者用集合做了缓存,并且只重写了了 unexport。
(三)DubboProtocol
该类是 dubbo 协议的核心实现,其中增加了比如延迟加载等处理。并且其中还包括了对服务暴露和服务引用的逻辑处理。
1. 属性
public static final String NAME = “dubbo”;

/**
* 默认端口号
*/
public static final int DEFAULT_PORT = 20880;
/**
* 回调名称
*/
private static final String IS_CALLBACK_SERVICE_INVOKE = “_isCallBackServiceInvoke”;
/**
* dubbo 协议的单例
*/
private static DubboProtocol INSTANCE;
/**
* 信息交换服务器集合 key:host:port value:ExchangeServer
*/
private final Map<String, ExchangeServer> serverMap = new ConcurrentHashMap<String, ExchangeServer>(); // <host:port,Exchanger>
/**
* 信息交换客户端集合
*/
private final Map<String, ReferenceCountExchangeClient> referenceClientMap = new ConcurrentHashMap<String, ReferenceCountExchangeClient>(); // <host:port,Exchanger>
/**
* 懒加载的客户端集合
*/
private final ConcurrentMap<String, LazyConnectExchangeClient> ghostClientMap = new ConcurrentHashMap<String, LazyConnectExchangeClient>();
/**
* 锁集合
*/
private final ConcurrentMap<String, Object> locks = new ConcurrentHashMap<String, Object>();
/**
* 序列化类名集合
*/
private final Set<String> optimizers = new ConcurrentHashSet<String>();
//consumer side export a stub service for dispatching event
//servicekey-stubmethods
/**
* 本地存根服务方法集合
*/
private final ConcurrentMap<String, String> stubServiceMethodsMap = new ConcurrentHashMap<String, String>();
/**
* 新建一个请求处理器
*/
private ExchangeHandler requestHandler = new ExchangeHandlerAdapter() {
/**
* 回复请求结果,返回的是请求结果
* @param channel
* @param message
* @return
* @throws RemotingException
*/
@Override
public Object reply(ExchangeChannel channel, Object message) throws RemotingException {
// 如果请求消息属于会话域
if (message instanceof Invocation) {
Invocation inv = (Invocation) message;
// 获得暴露的 invoker
Invoker<?> invoker = getInvoker(channel, inv);
// need to consider backward-compatibility if it’s a callback
// 如果是回调服务
if (Boolean.TRUE.toString().equals(inv.getAttachments().get(IS_CALLBACK_SERVICE_INVOKE))) {
// 获得 方法定义
String methodsStr = invoker.getUrl().getParameters().get(“methods”);
boolean hasMethod = false;
// 判断看是否有会话域中的方法
if (methodsStr == null || methodsStr.indexOf(“,”) == -1) {
hasMethod = inv.getMethodName().equals(methodsStr);
} else {
// 如果方法不止一个,则分割后遍历查询,找到了则设置为 true
String[] methods = methodsStr.split(“,”);
for (String method : methods) {
if (inv.getMethodName().equals(method)) {
hasMethod = true;
break;
}
}
}
// 如果没有该方法,则打印告警日志
if (!hasMethod) {
logger.warn(new IllegalStateException(“The methodName ” + inv.getMethodName()
+ ” not found in callback service interface ,invoke will be ignored.”
+ ” please update the api interface. url is:”
+ invoker.getUrl()) + ” ,invocation is :” + inv);
return null;
}
}
// 设置远程地址
RpcContext.getContext().setRemoteAddress(channel.getRemoteAddress());
// 调用下一个调用链
return invoker.invoke(inv);
}
// 否则抛出异常
throw new RemotingException(channel, “Unsupported request: ”
+ (message == null ? null : (message.getClass().getName() + “: ” + message))
+ “, channel: consumer: ” + channel.getRemoteAddress() + ” –> provider: ” + channel.getLocalAddress());
}

/**
* 接收消息
* @param channel
* @param message
* @throws RemotingException
*/
@Override
public void received(Channel channel, Object message) throws RemotingException {
// 如果消息是会话域中的消息,则调用 reply 方法。
if (message instanceof Invocation) {
reply((ExchangeChannel) channel, message);
} else {
super.received(channel, message);
}
}

@Override
public void connected(Channel channel) throws RemotingException {
// 接收连接事件
invoke(channel, Constants.ON_CONNECT_KEY);
}

@Override
public void disconnected(Channel channel) throws RemotingException {
if (logger.isInfoEnabled()) {
logger.info(“disconnected from ” + channel.getRemoteAddress() + “,url:” + channel.getUrl());
}
// 接收断开连接事件
invoke(channel, Constants.ON_DISCONNECT_KEY);
}

/**
* 接收事件
* @param channel
* @param methodKey
*/
private void invoke(Channel channel, String methodKey) {
// 创建会话域
Invocation invocation = createInvocation(channel, channel.getUrl(), methodKey);
if (invocation != null) {
try {
// 接收事件
received(channel, invocation);
} catch (Throwable t) {
logger.warn(“Failed to invoke event method ” + invocation.getMethodName() + “(), cause: ” + t.getMessage(), t);
}
}
}

/**
* 创建会话域, 把 url 内的值加入到会话域的附加值中
* @param channel
* @param url
* @param methodKey
* @return
*/
private Invocation createInvocation(Channel channel, URL url, String methodKey) {
// 获得方法,methodKey 是 onconnect 或者 ondisconnect
String method = url.getParameter(methodKey);
if (method == null || method.length() == 0) {
return null;
}
// 创建一个 rpc 会话域
RpcInvocation invocation = new RpcInvocation(method, new Class<?>[0], new Object[0]);
// 加入附加值 path
invocation.setAttachment(Constants.PATH_KEY, url.getPath());
// 加入附加值 group
invocation.setAttachment(Constants.GROUP_KEY, url.getParameter(Constants.GROUP_KEY));
// 加入附加值 interface
invocation.setAttachment(Constants.INTERFACE_KEY, url.getParameter(Constants.INTERFACE_KEY));
// 加入附加值 version
invocation.setAttachment(Constants.VERSION_KEY, url.getParameter(Constants.VERSION_KEY));
// 如果是本地存根服务,则加入附加值 dubbo.stub.event 为 true
if (url.getParameter(Constants.STUB_EVENT_KEY, false)) {
invocation.setAttachment(Constants.STUB_EVENT_KEY, Boolean.TRUE.toString());
}
return invocation;
}
};
该属性中关键的是实例化了一个请求处理器,其中实现了基于 dubbo 协议等连接、取消连接、回复请求结果等方法。
2.export
@Override
public <T> Exporter<T> export(Invoker<T> invoker) throws RpcException {
URL url = invoker.getUrl();

// export service.
// 得到服务 key group+”/”+serviceName+”:”+serviceVersion+”:”+port
String key = serviceKey(url);
// 创建 exporter
DubboExporter<T> exporter = new DubboExporter<T>(invoker, key, exporterMap);
// 加入到集合
exporterMap.put(key, exporter);

//export an stub service for dispatching event
Boolean isStubSupportEvent = url.getParameter(Constants.STUB_EVENT_KEY, Constants.DEFAULT_STUB_EVENT);
Boolean isCallbackservice = url.getParameter(Constants.IS_CALLBACK_SERVICE, false);
// 如果是本地存根事件而不是回调服务
if (isStubSupportEvent && !isCallbackservice) {
// 获得本地存根的方法
String stubServiceMethods = url.getParameter(Constants.STUB_EVENT_METHODS_KEY);
// 如果为空,则抛出异常
if (stubServiceMethods == null || stubServiceMethods.length() == 0) {
if (logger.isWarnEnabled()) {
logger.warn(new IllegalStateException(“consumer [” + url.getParameter(Constants.INTERFACE_KEY) +
“], has set stubproxy support event ,but no stub methods founded.”));
}
} else {
// 加入集合
stubServiceMethodsMap.put(url.getServiceKey(), stubServiceMethods);
}
}

// 打开服务
openServer(url);
// 序列化
optimizeSerialization(url);
return exporter;
}
该方法是基于 dubbo 协议的服务暴露,除了对于存根服务和本地服务进行标记以外,打开服务和序列化分别在 openServer 和 optimizeSerialization 中实现。
3.openServer
private void openServer(URL url) {
// find server.
String key = url.getAddress();
//client can export a service which’s only for server to invoke
// 客户端是否可以暴露仅供服务器调用的服务
boolean isServer = url.getParameter(Constants.IS_SERVER_KEY, true);
// 如果是的话
if (isServer) {
// 获得信息交换服务器
ExchangeServer server = serverMap.get(key);
if (server == null) {
// 重新创建服务器对象,然后放入集合
serverMap.put(key, createServer(url));
} else {
// server supports reset, use together with override
// 重置
server.reset(url);
}
}
}
该方法就是打开服务。其中的逻辑其实是把服务对象放入集合中进行缓存,如果该地址对应的服务器不存在,则调用 createServer 创建一个服务器对象。
4.createServer
private ExchangeServer createServer(URL url) {
// send readonly event when server closes, it’s enabled by default
// 服务器关闭时发送 readonly 事件,默认情况下启用
url = url.addParameterIfAbsent(Constants.CHANNEL_READONLYEVENT_SENT_KEY, Boolean.TRUE.toString());
// enable heartbeat by default
// 心跳默认间隔一分钟
url = url.addParameterIfAbsent(Constants.HEARTBEAT_KEY, String.valueOf(Constants.DEFAULT_HEARTBEAT));
// 获得远程通讯服务端实现方式,默认用 netty3
String str = url.getParameter(Constants.SERVER_KEY, Constants.DEFAULT_REMOTING_SERVER);

/**
* 如果没有该配置,则抛出异常
*/
if (str != null && str.length() > 0 && !ExtensionLoader.getExtensionLoader(Transporter.class).hasExtension(str))
throw new RpcException(“Unsupported server type: ” + str + “, url: ” + url);

/**
* 添加编解码器 DubboCodec 实现
*/
url = url.addParameter(Constants.CODEC_KEY, DubboCodec.NAME);
ExchangeServer server;
try {
// 启动服务器
server = Exchangers.bind(url, requestHandler);
} catch (RemotingException e) {
throw new RpcException(“Fail to start server(url: ” + url + “) ” + e.getMessage(), e);
}
// 获得客户端侧设置的远程通信方式
str = url.getParameter(Constants.CLIENT_KEY);
if (str != null && str.length() > 0) {
// 获得远程通信的实现集合
Set<String> supportedTypes = ExtensionLoader.getExtensionLoader(Transporter.class).getSupportedExtensions();
// 如果客户端侧设置的远程通信方式不在支持的方式中,则抛出异常
if (!supportedTypes.contains(str)) {
throw new RpcException(“Unsupported client type: ” + str);
}
}
return server;
}

该方法就是根据 url 携带的远程通信实现方法来创建一个服务器对象。
5.optimizeSerialization
private void optimizeSerialization(URL url) throws RpcException {
// 获得类名
String className = url.getParameter(Constants.OPTIMIZER_KEY, “”);
if (StringUtils.isEmpty(className) || optimizers.contains(className)) {
return;
}

logger.info(“Optimizing the serialization process for Kryo, FST, etc…”);

try {
// 加载类
Class clazz = Thread.currentThread().getContextClassLoader().loadClass(className);
if (!SerializationOptimizer.class.isAssignableFrom(clazz)) {
throw new RpcException(“The serialization optimizer ” + className + ” isn’t an instance of ” + SerializationOptimizer.class.getName());
}

// 强制类型转化为 SerializationOptimizer
SerializationOptimizer optimizer = (SerializationOptimizer) clazz.newInstance();

if (optimizer.getSerializableClasses() == null) {
return;
}

// 遍历序列化的类,把该类放入到集合进行缓存
for (Class c : optimizer.getSerializableClasses()) {
SerializableClassRegistry.registerClass(c);
}

// 加入到集合
optimizers.add(className);
} catch (ClassNotFoundException e) {
throw new RpcException(“Cannot find the serialization optimizer class: ” + className, e);
} catch (InstantiationException e) {
throw new RpcException(“Cannot instantiate the serialization optimizer class: ” + className, e);
} catch (IllegalAccessException e) {
throw new RpcException(“Cannot instantiate the serialization optimizer class: ” + className, e);
}
}

该方法是把序列化的类放入到集合,以便进行序列化
6.refer
@Override
public <T> Invoker<T> refer(Class<T> serviceType, URL url) throws RpcException {
// 序列化
optimizeSerialization(url);
// create rpc invoker. 创建一个 DubboInvoker 对象
DubboInvoker<T> invoker = new DubboInvoker<T>(serviceType, url, getClients(url), invokers);
// 把该 invoker 放入集合
invokers.add(invoker);
return invoker;
}

该方法是服务引用,其中就是新建一个 DubboInvoker 对象后把它放入到集合。
7.getClients
private ExchangeClient[] getClients(URL url) {
// whether to share connection
// 一个连接是否对于一个服务
boolean service_share_connect = false;
// 获得 url 中欢愉连接共享的配置 默认为 0
int connections = url.getParameter(Constants.CONNECTIONS_KEY, 0);
// if not configured, connection is shared, otherwise, one connection for one service
// 如果为 0,则是共享类,并且连接数为 1
if (connections == 0) {
service_share_connect = true;
connections = 1;
}

// 创建数组
ExchangeClient[] clients = new ExchangeClient[connections];
for (int i = 0; i < clients.length; i++) {
// 如果共享,则获得共享客户端对象,否则新建客户端
if (service_share_connect) {
clients[i] = getSharedClient(url);
} else {
clients[i] = initClient(url);
}
}
return clients;
}

该方法是获得客户端集合的方法,分为共享客户端和非共享客户端。共享客户端是共用同一个连接,非共享客户端是每个客户端都有自己的一个连接。
8.getSharedClient
private ExchangeClient getSharedClient(URL url) {
String key = url.getAddress();
// 从集合中取出客户端对象
ReferenceCountExchangeClient client = referenceClientMap.get(key);
// 如果不为空并且没关闭连接,则计数器加 1,返回
if (client != null) {
if (!client.isClosed()) {
client.incrementAndGetCount();
return client;
} else {
// 如果连接断开,则从集合中移除
referenceClientMap.remove(key);
}
}

locks.putIfAbsent(key, new Object());
synchronized (locks.get(key)) {
// 如果集合中有该 key
if (referenceClientMap.containsKey(key)) {
// 则直接返回 client
return referenceClientMap.get(key);
}

// 否则新建一个连接
ExchangeClient exchangeClient = initClient(url);
client = new ReferenceCountExchangeClient(exchangeClient, ghostClientMap);
// 存入集合
referenceClientMap.put(key, client);
// 从 ghostClientMap 中移除
ghostClientMap.remove(key);
// 从对象锁中移除
locks.remove(key);
return client;
}
}

该方法是获得分享的客户端连接。
9.initClient
private ExchangeClient initClient(URL url) {

// client type setting.
// 获得客户端的实现方法 默认 netty3
String str = url.getParameter(Constants.CLIENT_KEY, url.getParameter(Constants.SERVER_KEY, Constants.DEFAULT_REMOTING_CLIENT));

// 添加编码器
url = url.addParameter(Constants.CODEC_KEY, DubboCodec.NAME);
// enable heartbeat by default
// 默认开启心跳
url = url.addParameterIfAbsent(Constants.HEARTBEAT_KEY, String.valueOf(Constants.DEFAULT_HEARTBEAT));

// BIO is not allowed since it has severe performance issue.
if (str != null && str.length() > 0 && !ExtensionLoader.getExtensionLoader(Transporter.class).hasExtension(str)) {
throw new RpcException(“Unsupported client type: ” + str + “,” +
” supported client type is ” + StringUtils.join(ExtensionLoader.getExtensionLoader(Transporter.class).getSupportedExtensions(), ” “));
}

ExchangeClient client;
try {
// connection should be lazy
// 是否需要延迟连接,,默认不开启
if (url.getParameter(Constants.LAZY_CONNECT_KEY, false)) {
// 创建延迟连接的客户端
client = new LazyConnectExchangeClient(url, requestHandler);
} else {
// 否则就直接连接
client = Exchangers.connect(url, requestHandler);
}
} catch (RemotingException e) {
throw new RpcException(“Fail to create remoting client for service(” + url + “): ” + e.getMessage(), e);
}
return client;
}

该方法是新建一个客户端连接
10.destroy
@Override
public void destroy() {
// 遍历服务器逐个关闭
for (String key : new ArrayList<String>(serverMap.keySet())) {
ExchangeServer server = serverMap.remove(key);
if (server != null) {
try {
if (logger.isInfoEnabled()) {
logger.info(“Close dubbo server: ” + server.getLocalAddress());
}
server.close(ConfigUtils.getServerShutdownTimeout());
} catch (Throwable t) {
logger.warn(t.getMessage(), t);
}
}
}

// 遍历客户端集合逐个关闭
for (String key : new ArrayList<String>(referenceClientMap.keySet())) {
ExchangeClient client = referenceClientMap.remove(key);
if (client != null) {
try {
if (logger.isInfoEnabled()) {
logger.info(“Close dubbo connect: ” + client.getLocalAddress() + “–>” + client.getRemoteAddress());
}
client.close(ConfigUtils.getServerShutdownTimeout());
} catch (Throwable t) {
logger.warn(t.getMessage(), t);
}
}
}

// 遍历懒加载的集合,逐个关闭客户端
for (String key : new ArrayList<String>(ghostClientMap.keySet())) {
ExchangeClient client = ghostClientMap.remove(key);
if (client != null) {
try {
if (logger.isInfoEnabled()) {
logger.info(“Close dubbo connect: ” + client.getLocalAddress() + “–>” + client.getRemoteAddress());
}
client.close(ConfigUtils.getServerShutdownTimeout());
} catch (Throwable t) {
logger.warn(t.getMessage(), t);
}
}
}
stubServiceMethodsMap.clear();
super.destroy();
}

该方法是销毁的方法重写。
(四)ChannelWrappedInvoker
该类是对当前通道内的客户端调用消息进行包装
1. 属性
/**
* 通道
*/
private final Channel channel;
/**
* 服务 key
*/
private final String serviceKey;
/**
* 当前的客户端
*/
private final ExchangeClient currentClient;

2.doInvoke
@Override
protected Result doInvoke(Invocation invocation) throws Throwable {
RpcInvocation inv = (RpcInvocation) invocation;
// use interface’s name as service path to export if it’s not found on client side
// 设置服务 path,默认用接口名称
inv.setAttachment(Constants.PATH_KEY, getInterface().getName());
// 设置回调的服务 key
inv.setAttachment(Constants.CALLBACK_SERVICE_KEY, serviceKey);

try {
// 如果是异步的
if (getUrl().getMethodParameter(invocation.getMethodName(), Constants.ASYNC_KEY, false)) {// may have concurrency issue
// 直接发送请求消息
currentClient.send(inv, getUrl().getMethodParameter(invocation.getMethodName(), Constants.SENT_KEY, false));
return new RpcResult();
}
// 获得超时时间
int timeout = getUrl().getMethodParameter(invocation.getMethodName(), Constants.TIMEOUT_KEY, Constants.DEFAULT_TIMEOUT);
if (timeout > 0) {
return (Result) currentClient.request(inv, timeout).get();
} else {
return (Result) currentClient.request(inv).get();
}
} catch (RpcException e) {
throw e;
} catch (TimeoutException e) {
throw new RpcException(RpcException.TIMEOUT_EXCEPTION, e.getMessage(), e);
} catch (RemotingException e) {
throw new RpcException(RpcException.NETWORK_EXCEPTION, e.getMessage(), e);
} catch (Throwable e) {// here is non-biz exception, wrap it.
throw new RpcException(e.getMessage(), e);
}
}

该方法是在 invoker 调用的时候对发送请求消息进行了包装。
3.ChannelWrapper
该类是个内部没,继承了 ClientDelegate,其中将编码器变成了 dubbo 的编码器,其他方法比较简单。
(五)DecodeableRpcInvocation
该类主要做了对于会话域内的数据进行序列化和解码。
1. 属性
private static final Logger log = LoggerFactory.getLogger(DecodeableRpcInvocation.class);

/**
* 通道
*/
private Channel channel;

/**
* 序列化类型
*/
private byte serializationType;

/**
* 输入流
*/
private InputStream inputStream;

/**
* 请求
*/
private Request request;

/**
* 是否解码
*/
private volatile boolean hasDecoded;

2.decode
@Override
public void decode() throws Exception {
// 如果没有解码,则进行解码
if (!hasDecoded && channel != null && inputStream != null) {
try {
decode(channel, inputStream);
} catch (Throwable e) {
if (log.isWarnEnabled()) {
log.warn(“Decode rpc invocation failed: ” + e.getMessage(), e);
}
request.setBroken(true);
request.setData(e);
} finally {
// 设置已经解码
hasDecoded = true;
}
}
}

@Override
public Object decode(Channel channel, InputStream input) throws IOException {
// 对数据进行反序列化
ObjectInput in = CodecSupport.getSerialization(channel.getUrl(), serializationType)
.deserialize(channel.getUrl(), input);

// dubbo 版本
String dubboVersion = in.readUTF();
// 请求中放入 dubbo 版本
request.setVersion(dubboVersion);
// 附加值内加入 dubbo 版本,path 以及版本号
setAttachment(Constants.DUBBO_VERSION_KEY, dubboVersion);

setAttachment(Constants.PATH_KEY, in.readUTF());
setAttachment(Constants.VERSION_KEY, in.readUTF());

// 设置方法名称
setMethodName(in.readUTF());
try {
// 方法参数数组
Object[] args;
// 方法参数类型数组
Class<?>[] pts;
// 描述
String desc = in.readUTF();
// 如果为空,则方法参数数组和对方法参数类型数组都设置为空
if (desc.length() == 0) {
pts = DubboCodec.EMPTY_CLASS_ARRAY;
args = DubboCodec.EMPTY_OBJECT_ARRAY;
} else {
// 分割成类,获得类数组
pts = ReflectUtils.desc2classArray(desc);
// 创建等长等数组
args = new Object[pts.length];
for (int i = 0; i < args.length; i++) {
try {
// 读取对象放入数组中
args[i] = in.readObject(pts[i]);
} catch (Exception e) {
if (log.isWarnEnabled()) {
log.warn(“Decode argument failed: ” + e.getMessage(), e);
}
}
}
}
// 设置参数类型
setParameterTypes(pts);

Map<String, String> map = (Map<String, String>) in.readObject(Map.class);
if (map != null && map.size() > 0) {
// 获得所有附加值
Map<String, String> attachment = getAttachments();
if (attachment == null) {
attachment = new HashMap<String, String>();
}
// 把流中读到的配置放入附加值
attachment.putAll(map);
// 放回去
setAttachments(attachment);
}
//decode argument ,may be callback
for (int i = 0; i < args.length; i++) {
// 如果是回调,则再一次解码
args[i] = decodeInvocationArgument(channel, this, pts, i, args[i]);
}

setArguments(args);

} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read invocation data failed.”, e));
} finally {
if (in instanceof Cleanable) {
((Cleanable) in).cleanup();
}
}
return this;
}

该方法就是处理 Invocation 内数据的逻辑,其中主要是做了序列化和解码。把读取出来的设置放入对对应位置传递给后面的调用。
(六)DecodeableRpcResult
该类是做了基于 dubbo 协议对 prc 结果的解码
1. 属性
private static final Logger log = LoggerFactory.getLogger(DecodeableRpcResult.class);

/**
* 通道
*/
private Channel channel;

/**
* 序列化类型
*/
private byte serializationType;

/**
* 输入流
*/
private InputStream inputStream;

/**
* 响应
*/
private Response response;

/**
* 会话域
*/
private Invocation invocation;

/**
* 是否解码
*/
private volatile boolean hasDecoded;

2.decode
@Override
public Object decode(Channel channel, InputStream input) throws IOException {
// 反序列化
ObjectInput in = CodecSupport.getSerialization(channel.getUrl(), serializationType)
.deserialize(channel.getUrl(), input);

byte flag = in.readByte();
// 根据返回的不同结果来进行处理
switch (flag) {
case DubboCodec.RESPONSE_NULL_VALUE:
// 返回结果为空
break;
case DubboCodec.RESPONSE_VALUE:
//
try {
// 获得返回类型数组
Type[] returnType = RpcUtils.getReturnTypes(invocation);
// 根据返回类型读取返回结果并且放入 RpcResult
setValue(returnType == null || returnType.length == 0 ? in.readObject() :
(returnType.length == 1 ? in.readObject((Class<?>) returnType[0])
: in.readObject((Class<?>) returnType[0], returnType[1])));
} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read response data failed.”, e));
}
break;
case DubboCodec.RESPONSE_WITH_EXCEPTION:
// 返回结果有异常
try {
Object obj = in.readObject();
// 把异常放入 RpcResult
if (obj instanceof Throwable == false)
throw new IOException(“Response data error, expect Throwable, but get ” + obj);
setException((Throwable) obj);
} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read response data failed.”, e));
}
break;
case DubboCodec.RESPONSE_NULL_VALUE_WITH_ATTACHMENTS:
// 返回值为空,但是有附加值
try {
// 把附加值加入到 RpcResult
setAttachments((Map<String, String>) in.readObject(Map.class));
} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read response data failed.”, e));
}
break;
case DubboCodec.RESPONSE_VALUE_WITH_ATTACHMENTS:
// 返回值
try {
// 设置返回结果
Type[] returnType = RpcUtils.getReturnTypes(invocation);
setValue(returnType == null || returnType.length == 0 ? in.readObject() :
(returnType.length == 1 ? in.readObject((Class<?>) returnType[0])
: in.readObject((Class<?>) returnType[0], returnType[1])));
// 设置附加值
setAttachments((Map<String, String>) in.readObject(Map.class));
} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read response data failed.”, e));
}
break;
case DubboCodec.RESPONSE_WITH_EXCEPTION_WITH_ATTACHMENTS:
// 返回结果有异常并且有附加值
try {
// 设置异常
Object obj = in.readObject();
if (obj instanceof Throwable == false)
throw new IOException(“Response data error, expect Throwable, but get ” + obj);
setException((Throwable) obj);
// 设置附加值
setAttachments((Map<String, String>) in.readObject(Map.class));
} catch (ClassNotFoundException e) {
throw new IOException(StringUtils.toString(“Read response data failed.”, e));
}
break;
default:
throw new IOException(“Unknown result flag, expect ‘0’ ‘1’ ‘2’, get ” + flag);
}
if (in instanceof Cleanable) {
((Cleanable) in).cleanup();
}
return this;
}

@Override
public void decode() throws Exception {
// 如果没有解码
if (!hasDecoded && channel != null && inputStream != null) {
try {
// 进行解码
decode(channel, inputStream);
} catch (Throwable e) {
if (log.isWarnEnabled()) {
log.warn(“Decode rpc result failed: ” + e.getMessage(), e);
}
response.setStatus(Response.CLIENT_ERROR);
response.setErrorMessage(StringUtils.toString(e));
} finally {
hasDecoded = true;
}
}
}

该方法是对响应结果的解码,其中根据不同的返回结果来对 RpcResult 设置不同的值。
(七)LazyConnectExchangeClient
该类实现了 ExchangeClient 接口,是 ExchangeClient 的装饰器,用到了装饰模式,是延迟连接的客户端实现类。
1. 属性
// when this warning rises from invocation, program probably have bug.
/**
* 延迟连接请求错误 key
*/
static final String REQUEST_WITH_WARNING_KEY = “lazyclient_request_with_warning”;
private final static Logger logger = LoggerFactory.getLogger(LazyConnectExchangeClient.class);
/**
* 是否在延迟连接请求时错误
*/
protected final boolean requestWithWarning;
/**
* url 对象
*/
private final URL url;
/**
* 请求处理器
*/
private final ExchangeHandler requestHandler;
/**
* 连接锁
*/
private final Lock connectLock = new ReentrantLock();
// lazy connect, initial state for connection
/**
* 初始化状态
*/
private final boolean initialState;
/**
* 客户端对象
*/
private volatile ExchangeClient client;
/**
* 错误次数
*/
private AtomicLong warningcount = new AtomicLong(0);

可以看到有属性 ExchangeClient client,该类中很多方法就直接调用了 client 的方法。
2. 构造方法
public LazyConnectExchangeClient(URL url, ExchangeHandler requestHandler) {
// lazy connect, need set send.reconnect = true, to avoid channel bad status.
// 默认有重连
this.url = url.addParameter(Constants.SEND_RECONNECT_KEY, Boolean.TRUE.toString());
this.requestHandler = requestHandler;
// 默认延迟连接初始化成功
this.initialState = url.getParameter(Constants.LAZY_CONNECT_INITIAL_STATE_KEY, Constants.DEFAULT_LAZY_CONNECT_INITIAL_STATE);
// 默认没有错误
this.requestWithWarning = url.getParameter(REQUEST_WITH_WARNING_KEY, false);
}

3.initClient
private void initClient() throws RemotingException {
// 如果客户端已经初始化,则直接返回
if (client != null)
return;
if (logger.isInfoEnabled()) {
logger.info(“Lazy connect to ” + url);
}
// 获得连接锁
connectLock.lock();
try {
// 二次判空
if (client != null)
return;
// 新建一个客户端
this.client = Exchangers.connect(url, requestHandler);
} finally {
// 释放锁
connectLock.unlock();
}
}

该方法是初始化客户端的方法。
4.request
@Override
public ResponseFuture request(Object request) throws RemotingException {
warning(request);
initClient();
return client.request(request);
}

该方法在调用 client.request 前调用了前面两个方法,initClient 我在上面讲到了,就是用来初始化客户端的。而 warning 是用来报错的。
5.warning
private void warning(Object request) {
if (requestWithWarning) {
// 每 5000 次报错一次
if (warningcount.get() % 5000 == 0) {
logger.warn(new IllegalStateException(“safe guard client , should not be called ,must have a bug.”));
}
warningcount.incrementAndGet();
}
}

每 5000 次记录报错一次。
(八)ReferenceCountExchangeClient
该类也是对 ExchangeClient 的装饰,其中增强了调用次数多功能。
1. 属性
/**
* url 对象
*/
private final URL url;
/**
* 计数
*/
private final AtomicInteger refenceCount = new AtomicInteger(0);

// private final ExchangeHandler handler;
/**
* 延迟连接客户端集合
*/
private final ConcurrentMap<String, LazyConnectExchangeClient> ghostClientMap;
/**
* 客户端对象
*/
private ExchangeClient client;

2.replaceWithLazyClient
// ghost client
private LazyConnectExchangeClient replaceWithLazyClient() {
// this is a defensive operation to avoid client is closed by accident, the initial state of the client is false
// 设置延迟连接初始化状态、是否重连、是否已经重连等配置
URL lazyUrl = url.addParameter(Constants.LAZY_CONNECT_INITIAL_STATE_KEY, Boolean.FALSE)
.addParameter(Constants.RECONNECT_KEY, Boolean.FALSE)
.addParameter(Constants.SEND_RECONNECT_KEY, Boolean.TRUE.toString())
.addParameter(“warning”, Boolean.TRUE.toString())
.addParameter(LazyConnectExchangeClient.REQUEST_WITH_WARNING_KEY, true)
.addParameter(“_client_memo”, “referencecounthandler.replacewithlazyclient”);

// 获得服务地址
String key = url.getAddress();
// in worst case there’s only one ghost connection.
// 从集合中获取客户端
LazyConnectExchangeClient gclient = ghostClientMap.get(key);
// 如果对应等客户端不存在或者已经关闭连接,则重新创建一个延迟连接等客户端,并且放入集合
if (gclient == null || gclient.isClosed()) {
gclient = new LazyConnectExchangeClient(lazyUrl, client.getExchangeHandler());
ghostClientMap.put(key, gclient);
}
return gclient;
}

该方法是用延迟连接替代,该方法在 close 方法中被调用。
3.close
@Override
public void close(int timeout) {
if (refenceCount.decrementAndGet() <= 0) {
if (timeout == 0) {
client.close();
} else {
client.close(timeout);
}
client = replaceWithLazyClient();
}
}

(九)FutureAdapter
该类实现了 Future 接口,是响应的 Future 适配器。其中是基于 ResponseFuture 做适配。其中比较简单,我就不多讲解了。
(十)CallbackServiceCodec
该类是针对回调服务的编解码器。
1. 属性
/**
* 代理工厂
*/
private static final ProxyFactory proxyFactory = ExtensionLoader.getExtensionLoader(ProxyFactory.class).getAdaptiveExtension();
/**
* dubbo 协议
*/
private static final DubboProtocol protocol = DubboProtocol.getDubboProtocol();
/**
* 回调的标志
*/
private static final byte CALLBACK_NONE = 0x0;
/**
* 回调的创建标志
*/
private static final byte CALLBACK_CREATE = 0x1;
/**
* 回调的销毁标志
*/
private static final byte CALLBACK_DESTROY = 0x2;
/**
* 回调参数 key
*/
private static final String INV_ATT_CALLBACK_KEY = “sys_callback_arg-“;

2.encodeInvocationArgument
public static Object encodeInvocationArgument(Channel channel, RpcInvocation inv, int paraIndex) throws IOException {
// get URL directly
// 直接获得 url
URL url = inv.getInvoker() == null ? null : inv.getInvoker().getUrl();
// 设置回调标志
byte callbackstatus = isCallBack(url, inv.getMethodName(), paraIndex);
// 获得参数集合
Object[] args = inv.getArguments();
// 获得参数类型集合
Class<?>[] pts = inv.getParameterTypes();
// 根据不同的回调状态来设置附加值和返回参数
switch (callbackstatus) {
case CallbackServiceCodec.CALLBACK_NONE:
return args[paraIndex];
case CallbackServiceCodec.CALLBACK_CREATE:
inv.setAttachment(INV_ATT_CALLBACK_KEY + paraIndex, exportOrunexportCallbackService(channel, url, pts[paraIndex], args[paraIndex], true));
return null;
case CallbackServiceCodec.CALLBACK_DESTROY:
inv.setAttachment(INV_ATT_CALLBACK_KEY + paraIndex, exportOrunexportCallbackService(channel, url, pts[paraIndex], args[paraIndex], false));
return null;
default:
return args[paraIndex];
}
}

该方法是对会话域的信息进行编码。
3.decodeInvocationArgument
public static Object decodeInvocationArgument(Channel channel, RpcInvocation inv, Class<?>[] pts, int paraIndex, Object inObject) throws IOException {
// if it’s a callback, create proxy on client side, callback interface on client side can be invoked through channel
// need get URL from channel and env when decode
URL url = null;
try {
// 获得 url
url = DubboProtocol.getDubboProtocol().getInvoker(channel, inv).getUrl();
} catch (RemotingException e) {
if (logger.isInfoEnabled()) {
logger.info(e.getMessage(), e);
}
return inObject;
}
// 获得回调状态
byte callbackstatus = isCallBack(url, inv.getMethodName(), paraIndex);
// 根据回调状态来返回结果
switch (callbackstatus) {
case CallbackServiceCodec.CALLBACK_NONE:
return inObject;
case CallbackServiceCodec.CALLBACK_CREATE:
try {
return referOrdestroyCallbackService(channel, url, pts[paraIndex], inv, Integer.parseInt(inv.getAttachment(INV_ATT_CALLBACK_KEY + paraIndex)), true);
} catch (Exception e) {
logger.error(e.getMessage(), e);
throw new IOException(StringUtils.toString(e));
}
case CallbackServiceCodec.CALLBACK_DESTROY:
try {
return referOrdestroyCallbackService(channel, url, pts[paraIndex], inv, Integer.parseInt(inv.getAttachment(INV_ATT_CALLBACK_KEY + paraIndex)), false);
} catch (Exception e) {
throw new IOException(StringUtils.toString(e));
}
default:
return inObject;
}
}

该方法是对会话域内的信息进行解码。
4.isCallBack
private static byte isCallBack(URL url, String methodName, int argIndex) {
// parameter callback rule: method-name.parameter-index(starting from 0).callback
// 参数的规则:ethod-name.parameter-index(starting from 0).callback
byte isCallback = CALLBACK_NONE;
if (url != null) {
// 获得回调的值
String callback = url.getParameter(methodName + “.” + argIndex + “.callback”);
if (callback != null) {
// 如果为 true,则设置为创建标志
if (callback.equalsIgnoreCase(“true”)) {
isCallback = CALLBACK_CREATE;
// 如果为 false,则设置为销毁标志
} else if (callback.equalsIgnoreCase(“false”)) {
isCallback = CALLBACK_DESTROY;
}
}
}
return isCallback;
}

该方法是根据 url 携带的参数设置回调的标志,以供执行不同的编解码逻辑。
5.exportOrunexportCallbackService
private static String exportOrunexportCallbackService(Channel channel, URL url, Class clazz, Object inst, Boolean export) throws IOException {
// 返回对象的 hashCode
int instid = System.identityHashCode(inst);

Map<String, String> params = new HashMap<String, String>(3);
// no need to new client again
// 设置不是服务端标志为否
params.put(Constants.IS_SERVER_KEY, Boolean.FALSE.toString());
// mark it’s a callback, for troubleshooting
// 设置是回调服务标志为 true
params.put(Constants.IS_CALLBACK_SERVICE, Boolean.TRUE.toString());
String group = url.getParameter(Constants.GROUP_KEY);
if (group != null && group.length() > 0) {
// 设置是消费侧还是提供侧
params.put(Constants.GROUP_KEY, group);
}
// add method, for verifying against method, automatic fallback (see dubbo protocol)
// 添加方法,在 dubbo 的协议里面用到
params.put(Constants.METHODS_KEY, StringUtils.join(Wrapper.getWrapper(clazz).getDeclaredMethodNames(), “,”));

Map<String, String> tmpmap = new HashMap<String, String>(url.getParameters());
tmpmap.putAll(params);
// 移除版本信息
tmpmap.remove(Constants.VERSION_KEY);// doesn’t need to distinguish version for callback
// 设置接口名
tmpmap.put(Constants.INTERFACE_KEY, clazz.getName());
// 创建服务暴露的 url
URL exporturl = new URL(DubboProtocol.NAME, channel.getLocalAddress().getAddress().getHostAddress(), channel.getLocalAddress().getPort(), clazz.getName() + “.” + instid, tmpmap);

// no need to generate multiple exporters for different channel in the same JVM, cache key cannot collide.

// 获得缓存的 key
String cacheKey = getClientSideCallbackServiceCacheKey(instid);
// 获得计数的 key
String countkey = getClientSideCountKey(clazz.getName());
// 如果是暴露服务
if (export) {
// one channel can have multiple callback instances, no need to re-export for different instance.
if (!channel.hasAttribute(cacheKey)) {
if (!isInstancesOverLimit(channel, url, clazz.getName(), instid, false)) {
// 获得代理对象
Invoker<?> invoker = proxyFactory.getInvoker(inst, clazz, exporturl);
// should destroy resource?
// 暴露服务
Exporter<?> exporter = protocol.export(invoker);
// this is used for tracing if instid has published service or not.
// 放到通道
channel.setAttribute(cacheKey, exporter);
logger.info(“export a callback service :” + exporturl + “, on ” + channel + “, url is: ” + url);
// 计数器加 1
increaseInstanceCount(channel, countkey);
}
}
} else {
// 如果通道内已经有该服务的缓存
if (channel.hasAttribute(cacheKey)) {
// 则获得该暴露者
Exporter<?> exporter = (Exporter<?>) channel.getAttribute(cacheKey);
// 取消暴露
exporter.unexport();
// 移除该缓存
channel.removeAttribute(cacheKey);
// 计数器减 1
decreaseInstanceCount(channel, countkey);
}
}
return String.valueOf(instid);
}

该方法是在客户端侧暴露服务和取消暴露服务。
6.referOrdestroyCallbackService
private static Object referOrdestroyCallbackService(Channel channel, URL url, Class<?> clazz, Invocation inv, int instid, boolean isRefer) {
Object proxy = null;
// 获得服务调用的缓存 key
String invokerCacheKey = getServerSideCallbackInvokerCacheKey(channel, clazz.getName(), instid);
// 获得代理缓存 key
String proxyCacheKey = getServerSideCallbackServiceCacheKey(channel, clazz.getName(), instid);
// 从通道内获得代理对象
proxy = channel.getAttribute(proxyCacheKey);
// 获得计数器 key
String countkey = getServerSideCountKey(channel, clazz.getName());
// 如果是服务引用
if (isRefer) {
// 如果代理对象为空
if (proxy == null) {
// 获得服务引用的 url
URL referurl = URL.valueOf(“callback://” + url.getAddress() + “/” + clazz.getName() + “?” + Constants.INTERFACE_KEY + “=” + clazz.getName());
referurl = referurl.addParametersIfAbsent(url.getParameters()).removeParameter(Constants.METHODS_KEY);
if (!isInstancesOverLimit(channel, referurl, clazz.getName(), instid, true)) {
@SuppressWarnings(“rawtypes”)
Invoker<?> invoker = new ChannelWrappedInvoker(clazz, channel, referurl, String.valueOf(instid));
// 获得代理类
proxy = proxyFactory.getProxy(invoker);
// 设置代理类
channel.setAttribute(proxyCacheKey, proxy);
// 设置实体域
channel.setAttribute(invokerCacheKey, invoker);
// 计数器加 1
increaseInstanceCount(channel, countkey);

//convert error fail fast .
//ignore concurrent problem.
Set<Invoker<?>> callbackInvokers = (Set<Invoker<?>>) channel.getAttribute(Constants.CHANNEL_CALLBACK_KEY);
if (callbackInvokers == null) {
// 创建回调的服务实体域集合
callbackInvokers = new ConcurrentHashSet<Invoker<?>>(1);
// 把该实体域加入集合中
callbackInvokers.add(invoker);
channel.setAttribute(Constants.CHANNEL_CALLBACK_KEY, callbackInvokers);
}
logger.info(“method ” + inv.getMethodName() + ” include a callback service :” + invoker.getUrl() + “, a proxy :” + invoker + ” has been created.”);
}
}
} else {
// 销毁
if (proxy != null) {
Invoker<?> invoker = (Invoker<?>) channel.getAttribute(invokerCacheKey);
try {
Set<Invoker<?>> callbackInvokers = (Set<Invoker<?>>) channel.getAttribute(Constants.CHANNEL_CALLBACK_KEY);
if (callbackInvokers != null) {
// 从集合中移除
callbackInvokers.remove(invoker);
}
// 销毁该调用
invoker.destroy();
} catch (Exception e) {
logger.error(e.getMessage(), e);
}
// cancel refer, directly remove from the map
// 取消引用,直接从集合中移除
channel.removeAttribute(proxyCacheKey);
channel.removeAttribute(invokerCacheKey);
// 计数器减 1
decreaseInstanceCount(channel, countkey);
}
}
return proxy;
}

该方法是在服务端侧进行服务引用或者销毁回调服务。
(十一)DubboCodec
该类是 dubbo 的编解码器,分别针对 dubbo 协议的 request 和 response 进行编码和解码。
1. 属性
/**
* dubbo 名称
*/
public static final String NAME = “dubbo”;
/**
* 协议版本号
*/
public static final String DUBBO_VERSION = Version.getProtocolVersion();
/**
* 响应携带着异常
*/
public static final byte RESPONSE_WITH_EXCEPTION = 0;
/**
* 响应
*/
public static final byte RESPONSE_VALUE = 1;
/**
* 响应结果为空
*/
public static final byte RESPONSE_NULL_VALUE = 2;
/**
* 响应结果有异常并且带有附加值
*/
public static final byte RESPONSE_WITH_EXCEPTION_WITH_ATTACHMENTS = 3;
/**
* 响应结果有附加值
*/
public static final byte RESPONSE_VALUE_WITH_ATTACHMENTS = 4;
/**
* 响应结果为空并带有附加值
*/
public static final byte RESPONSE_NULL_VALUE_WITH_ATTACHMENTS = 5;
/**
* 对象空集合
*/
public static final Object[] EMPTY_OBJECT_ARRAY = new Object[0];
/**
* 空的类集合
*/
public static final Class<?>[] EMPTY_CLASS_ARRAY = new Class<?>[0];
private static final Logger log = LoggerFactory.getLogger(DubboCodec.class);

2.decodeBody
@Override
protected Object decodeBody(Channel channel, InputStream is, byte[] header) throws IOException {
byte flag = header[2], proto = (byte) (flag & SERIALIZATION_MASK);
// get request id.
long id = Bytes.bytes2long(header, 4);
// 如果是 response
if ((flag & FLAG_REQUEST) == 0) {
// decode response.
// 创建一个 response
Response res = new Response(id);
// 如果是事件,则设置事件,这里有个问题,我提交了 pr 在新版本已经修复
if ((flag & FLAG_EVENT) != 0) {
res.setEvent(Response.HEARTBEAT_EVENT);
}
// get status.
// 设置状态
byte status = header[3];
res.setStatus(status);
try {
// 反序列化
ObjectInput in = CodecSupport.deserialize(channel.getUrl(), is, proto);
// 如果状态是响应成功
if (status == Response.OK) {
Object data;
// 如果是心跳事件,则按照心跳事件解码
if (res.isHeartbeat()) {
data = decodeHeartbeatData(channel, in);
} else if (res.isEvent()) {
// 如果是事件,则
data = decodeEventData(channel, in);
} else {
// 否则对结果进行解码
DecodeableRpcResult result;
if (channel.getUrl().getParameter(
Constants.DECODE_IN_IO_THREAD_KEY,
Constants.DEFAULT_DECODE_IN_IO_THREAD)) {
result = new DecodeableRpcResult(channel, res, is,
(Invocation) getRequestData(id), proto);
result.decode();
} else {
result = new DecodeableRpcResult(channel, res,
new UnsafeByteArrayInputStream(readMessageData(is)),
(Invocation) getRequestData(id), proto);
}
data = result;
}
// 把结果重新放入 response 中
res.setResult(data);
} else {
// 否则设置错误信息
res.setErrorMessage(in.readUTF());
}
} catch (Throwable t) {
if (log.isWarnEnabled()) {
log.warn(“Decode response failed: ” + t.getMessage(), t);
}
res.setStatus(Response.CLIENT_ERROR);
res.setErrorMessage(StringUtils.toString(t));
}
return res;
} else {
// decode request.
// 如果该消息是 request
Request req = new Request(id);
// 设置版本
req.setVersion(Version.getProtocolVersion());
// 设置是否是双向请求
req.setTwoWay((flag & FLAG_TWOWAY) != 0);
// 设置是否是事件,该地方问题也在新版本修复
if ((flag & FLAG_EVENT) != 0) {
req.setEvent(Request.HEARTBEAT_EVENT);
}
try {
Object data;
// 反序列化
ObjectInput in = CodecSupport.deserialize(channel.getUrl(), is, proto);
// 进行解码
if (req.isHeartbeat()) {
data = decodeHeartbeatData(channel, in);
} else if (req.isEvent()) {
data = decodeEventData(channel, in);
} else {
DecodeableRpcInvocation inv;
if (channel.getUrl().getParameter(
Constants.DECODE_IN_IO_THREAD_KEY,
Constants.DEFAULT_DECODE_IN_IO_THREAD)) {
inv = new DecodeableRpcInvocation(channel, req, is, proto);
inv.decode();
} else {
inv = new DecodeableRpcInvocation(channel, req,
new UnsafeByteArrayInputStream(readMessageData(is)), proto);
}
data = inv;
}
// 把 body 数据设置到 response
req.setData(data);
} catch (Throwable t) {
if (log.isWarnEnabled()) {
log.warn(“Decode request failed: ” + t.getMessage(), t);
}
// bad request
req.setBroken(true);
req.setData(t);
}
return req;
}
}

该方法是对 request 和 response 进行解码,用位运算来进行解码,其中的逻辑跟我在《dubbo 源码解析(十)远程通信——Exchange 层》中讲到的编解码器逻辑差不多。
3.encodeRequestData
@Override
protected void encodeRequestData(Channel channel, ObjectOutput out, Object data, String version) throws IOException {
RpcInvocation inv = (RpcInvocation) data;

// 输出版本
out.writeUTF(version);
// 输出 path
out.writeUTF(inv.getAttachment(Constants.PATH_KEY));
// 输出版本号
out.writeUTF(inv.getAttachment(Constants.VERSION_KEY));

// 输出方法名称
out.writeUTF(inv.getMethodName());
// 输出参数类型
out.writeUTF(ReflectUtils.getDesc(inv.getParameterTypes()));
// 输出参数
Object[] args = inv.getArguments();
if (args != null)
for (int i = 0; i < args.length; i++) {
out.writeObject(encodeInvocationArgument(channel, inv, i));
}
// 输出附加值
out.writeObject(inv.getAttachments());
}

该方法是对请求数据的编码。
4.encodeResponseData
@Override
protected void encodeResponseData(Channel channel, ObjectOutput out, Object data, String version) throws IOException {
Result result = (Result) data;
// currently, the version value in Response records the version of Request
boolean attach = Version.isSupportResponseAttatchment(version);
// 获得异常
Throwable th = result.getException();
if (th == null) {
Object ret = result.getValue();
// 根据结果的不同输出不同的值
if (ret == null) {
out.writeByte(attach ? RESPONSE_NULL_VALUE_WITH_ATTACHMENTS : RESPONSE_NULL_VALUE);
} else {
out.writeByte(attach ? RESPONSE_VALUE_WITH_ATTACHMENTS : RESPONSE_VALUE);
out.writeObject(ret);
}
} else {
// 如果有异常,则输出异常
out.writeByte(attach ? RESPONSE_WITH_EXCEPTION_WITH_ATTACHMENTS : RESPONSE_WITH_EXCEPTION);
out.writeObject(th);
}

if (attach) {
// returns current version of Response to consumer side.
// 在附加值中加入版本号
result.getAttachments().put(Constants.DUBBO_VERSION_KEY, Version.getProtocolVersion());
// 输出版本号
out.writeObject(result.getAttachments());
}
}

该方法是对响应数据的编码。
(十二)DubboCountCodec
该类是对 DubboCodec 的功能增强,增加了消息长度的限制。
public final class DubboCountCodec implements Codec2 {

private DubboCodec codec = new DubboCodec();

@Override
public void encode(Channel channel, ChannelBuffer buffer, Object msg) throws IOException {
codec.encode(channel, buffer, msg);
}

@Override
public Object decode(Channel channel, ChannelBuffer buffer) throws IOException {
// 保存读取的标志
int save = buffer.readerIndex();
MultiMessage result = MultiMessage.create();
do {
Object obj = codec.decode(channel, buffer);
// 粘包拆包
if (Codec2.DecodeResult.NEED_MORE_INPUT == obj) {
buffer.readerIndex(save);
break;
} else {
// 增加消息
result.addMessage(obj);
// 记录消息长度
logMessageLength(obj, buffer.readerIndex() – save);
save = buffer.readerIndex();
}
} while (true);
// 如果结果为空,则返回需要更多的输入
if (result.isEmpty()) {
return Codec2.DecodeResult.NEED_MORE_INPUT;
}
if (result.size() == 1) {
return result.get(0);
}
return result;
}

private void logMessageLength(Object result, int bytes) {
if (bytes <= 0) {
return;
}
// 如果是 request 类型
if (result instanceof Request) {
try {
// 设置附加值
((RpcInvocation) ((Request) result).getData()).setAttachment(
Constants.INPUT_KEY, String.valueOf(bytes));
} catch (Throwable e) {
/* ignore */
}
} else if (result instanceof Response) {
try {
// 设置附加值 输出的长度
((RpcResult) ((Response) result).getResult()).setAttachment(
Constants.OUTPUT_KEY, String.valueOf(bytes));
} catch (Throwable e) {
/* ignore */
}
}
}

}

(十三)TraceFilter
该过滤器是增强的功能是通道的跟踪,会在通道内把最大的调用次数和现在的调用数量放进去。方便使用 telnet 来跟踪服务的调用次数等。
1. 属性
/**
* 跟踪数量的最大值 key
*/
private static final String TRACE_MAX = “trace.max”;

/**
* 跟踪的数量
*/
private static final String TRACE_COUNT = “trace.count”;

/**
* 通道集合
*/
private static final ConcurrentMap<String, Set<Channel>> tracers = new ConcurrentHashMap<String, Set<Channel>>();

2.addTracer
public static void addTracer(Class<?> type, String method, Channel channel, int max) {
// 设置最大的数量
channel.setAttribute(TRACE_MAX, max);
// 设置当前的数量
channel.setAttribute(TRACE_COUNT, new AtomicInteger());
// 获得 key
String key = method != null && method.length() > 0 ? type.getName() + “.” + method : type.getName();
// 获得通道集合
Set<Channel> channels = tracers.get(key);
// 如果为空,则新建
if (channels == null) {
tracers.putIfAbsent(key, new ConcurrentHashSet<Channel>());
channels = tracers.get(key);
}
channels.add(channel);
}

该方法是对某一个通道进行跟踪,把现在的调用数量放到属性里面
3.removeTracer
public static void removeTracer(Class<?> type, String method, Channel channel) {
// 移除最大值属性
channel.removeAttribute(TRACE_MAX);
// 移除数量属性
channel.removeAttribute(TRACE_COUNT);
String key = method != null && method.length() > 0 ? type.getName() + “.” + method : type.getName();
Set<Channel> channels = tracers.get(key);
if (channels != null) {
// 集合中移除该通道
channels.remove(channel);
}
}

该方法是移除通道的跟踪。
4.invoke
@Override
public Result invoke(Invoker<?> invoker, Invocation invocation) throws RpcException {
// 开始时间
long start = System.currentTimeMillis();
// 调用下一个调用链 获得结果
Result result = invoker.invoke(invocation);
// 调用结束时间
long end = System.currentTimeMillis();
// 如果通道跟踪大小大于 0
if (tracers.size() > 0) {
// 服务 key
String key = invoker.getInterface().getName() + “.” + invocation.getMethodName();
// 获得通道集合
Set<Channel> channels = tracers.get(key);
if (channels == null || channels.isEmpty()) {
key = invoker.getInterface().getName();
channels = tracers.get(key);
}
if (channels != null && !channels.isEmpty()) {
// 遍历通道集合
for (Channel channel : new ArrayList<Channel>(channels)) {
// 如果通道是连接的
if (channel.isConnected()) {
try {
// 获得跟踪的最大数
int max = 1;
Integer m = (Integer) channel.getAttribute(TRACE_MAX);
if (m != null) {
max = (int) m;
}
// 获得跟踪数量
int count = 0;
AtomicInteger c = (AtomicInteger) channel.getAttribute(TRACE_COUNT);
if (c == null) {
c = new AtomicInteger();
channel.setAttribute(TRACE_COUNT, c);
}
count = c.getAndIncrement();
// 如果数量小于最大数量则发送
if (count < max) {
String prompt = channel.getUrl().getParameter(Constants.PROMPT_KEY, Constants.DEFAULT_PROMPT);
channel.send(“\r\n” + RpcContext.getContext().getRemoteAddress() + ” -> ”
+ invoker.getInterface().getName()
+ “.” + invocation.getMethodName()
+ “(” + JSON.toJSONString(invocation.getArguments()) + “)” + ” -> ” + JSON.toJSONString(result.getValue())
+ “\r\nelapsed: ” + (end – start) + ” ms.”
+ “\r\n\r\n” + prompt);
}
// 如果数量大于等于 max – 1,则移除该通道
if (count >= max – 1) {
channels.remove(channel);
}
} catch (Throwable e) {
channels.remove(channel);
logger.warn(e.getMessage(), e);
}
} else {
// 如果未连接,也移除该通道
channels.remove(channel);
}
}
}
}
return result;
}

该方法是当服务被调用时,进行跟踪或者取消跟踪的处理逻辑,是核心的功能增强逻辑。
(十四)FutureFilter
该类是处理异步和同步调用结果的过滤器。
1.invoke
@Override
public Result invoke(final Invoker<?> invoker, final Invocation invocation) throws RpcException {
// 是否是异步的调用
final boolean isAsync = RpcUtils.isAsync(invoker.getUrl(), invocation);

fireInvokeCallback(invoker, invocation);
// need to configure if there’s return value before the invocation in order to help invoker to judge if it’s
// necessary to return future.
Result result = invoker.invoke(invocation);
if (isAsync) {
// 调用异步处理
asyncCallback(invoker, invocation);
} else {
// 调用同步结果处理
syncCallback(invoker, invocation, result);
}
return result;
}

该方法中根据是否为异步调用来分别执行 asyncCallback 和 syncCallback 方法。
2.syncCallback
private void syncCallback(final Invoker<?> invoker, final Invocation invocation, final Result result) {
// 如果有异常
if (result.hasException()) {
// 则调用异常的结果处理
fireThrowCallback(invoker, invocation, result.getException());
} else {
// 调用正常的结果处理
fireReturnCallback(invoker, invocation, result.getValue());
}
}

该方法是同步调用的返回结果处理,比较简单。
3.asyncCallback
private void asyncCallback(final Invoker<?> invoker, final Invocation invocation) {
Future<?> f = RpcContext.getContext().getFuture();
if (f instanceof FutureAdapter) {
ResponseFuture future = ((FutureAdapter<?>) f).getFuture();
// 设置回调
future.setCallback(new ResponseCallback() {
@Override
public void done(Object rpcResult) {
// 如果结果为空,则打印错误日志
if (rpcResult == null) {
logger.error(new IllegalStateException(“invalid result value : null, expected ” + Result.class.getName()));
return;
}
///must be rpcResult
// 如果不是 Result 则打印错误日志
if (!(rpcResult instanceof Result)) {
logger.error(new IllegalStateException(“invalid result type :” + rpcResult.getClass() + “, expected ” + Result.class.getName()));
return;
}
Result result = (Result) rpcResult;
if (result.hasException()) {
// 如果有异常,则调用异常处理方法
fireThrowCallback(invoker, invocation, result.getException());
} else {
// 如果正常的返回结果,则调用正常的处理方法
fireReturnCallback(invoker, invocation, result.getValue());
}
}

@Override
public void caught(Throwable exception) {
fireThrowCallback(invoker, invocation, exception);
}
});
}
}

该方法是异步调用的结果处理,把异步返回结果的逻辑写在回调函数里面。
4.fireInvokeCallback
private void fireInvokeCallback(final Invoker<?> invoker, final Invocation invocation) {
// 获得调用的方法
final Method onInvokeMethod = (Method) StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_INVOKE_METHOD_KEY));
// 获得调用的服务
final Object onInvokeInst = StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_INVOKE_INSTANCE_KEY));

if (onInvokeMethod == null && onInvokeInst == null) {
return;
}
if (onInvokeMethod == null || onInvokeInst == null) {
throw new IllegalStateException(“service:” + invoker.getUrl().getServiceKey() + ” has a onreturn callback config , but no such ” + (onInvokeMethod == null ? “method” : “instance”) + ” found. url:” + invoker.getUrl());
}
// 如果不可以访问,则设置为可访问
if (!onInvokeMethod.isAccessible()) {
onInvokeMethod.setAccessible(true);
}

// 获得参数数组
Object[] params = invocation.getArguments();
try {
// 调用方法
onInvokeMethod.invoke(onInvokeInst, params);
} catch (InvocationTargetException e) {
fireThrowCallback(invoker, invocation, e.getTargetException());
} catch (Throwable e) {
fireThrowCallback(invoker, invocation, e);
}
}

该方法是调用方法的执行。
5.fireReturnCallback
private void fireReturnCallback(final Invoker<?> invoker, final Invocation invocation, final Object result) {
final Method onReturnMethod = (Method) StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_RETURN_METHOD_KEY));
final Object onReturnInst = StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_RETURN_INSTANCE_KEY));

//not set onreturn callback
if (onReturnMethod == null && onReturnInst == null) {
return;
}

if (onReturnMethod == null || onReturnInst == null) {
throw new IllegalStateException(“service:” + invoker.getUrl().getServiceKey() + ” has a onreturn callback config , but no such ” + (onReturnMethod == null ? “method” : “instance”) + ” found. url:” + invoker.getUrl());
}
if (!onReturnMethod.isAccessible()) {
onReturnMethod.setAccessible(true);
}

Object[] args = invocation.getArguments();
Object[] params;
// 获得返回结果类型
Class<?>[] rParaTypes = onReturnMethod.getParameterTypes();
// 设置参数和返回结果
if (rParaTypes.length > 1) {
if (rParaTypes.length == 2 && rParaTypes[1].isAssignableFrom(Object[].class)) {
params = new Object[2];
params[0] = result;
params[1] = args;
} else {
params = new Object[args.length + 1];
params[0] = result;
System.arraycopy(args, 0, params, 1, args.length);
}
} else {
params = new Object[]{result};
}
try {
// 调用方法
onReturnMethod.invoke(onReturnInst, params);
} catch (InvocationTargetException e) {
fireThrowCallback(invoker, invocation, e.getTargetException());
} catch (Throwable e) {
fireThrowCallback(invoker, invocation, e);
}
}

该方法是正常的返回结果的处理。
6.fireThrowCallback
private void fireThrowCallback(final Invoker<?> invoker, final Invocation invocation, final Throwable exception) {
final Method onthrowMethod = (Method) StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_THROW_METHOD_KEY));
final Object onthrowInst = StaticContext.getSystemContext().get(StaticContext.getKey(invoker.getUrl(), invocation.getMethodName(), Constants.ON_THROW_INSTANCE_KEY));

//onthrow callback not configured
if (onthrowMethod == null && onthrowInst == null) {
return;
}
if (onthrowMethod == null || onthrowInst == null) {
throw new IllegalStateException(“service:” + invoker.getUrl().getServiceKey() + ” has a onthrow callback config , but no such ” + (onthrowMethod == null ? “method” : “instance”) + ” found. url:” + invoker.getUrl());
}
if (!onthrowMethod.isAccessible()) {
onthrowMethod.setAccessible(true);
}
// 获得抛出异常的类型
Class<?>[] rParaTypes = onthrowMethod.getParameterTypes();
if (rParaTypes[0].isAssignableFrom(exception.getClass())) {
try {
Object[] args = invocation.getArguments();
Object[] params;

// 把类型和抛出的异常值放入返回结果
if (rParaTypes.length > 1) {
if (rParaTypes.length == 2 && rParaTypes[1].isAssignableFrom(Object[].class)) {
params = new Object[2];
params[0] = exception;
params[1] = args;
} else {
params = new Object[args.length + 1];
params[0] = exception;
System.arraycopy(args, 0, params, 1, args.length);
}
} else {
params = new Object[]{exception};
}
// 调用下一个调用连
onthrowMethod.invoke(onthrowInst, params);
} catch (Throwable e) {
logger.error(invocation.getMethodName() + “.call back method invoke error . callback method :” + onthrowMethod + “, url:” + invoker.getUrl(), e);
}
} else {
logger.error(invocation.getMethodName() + “.call back method invoke error . callback method :” + onthrowMethod + “, url:” + invoker.getUrl(), exception);
}
}

该方法是异常抛出时的结果处理。
(十五)ServerStatusChecker
该类是对于服务状态的监控设置。
public class ServerStatusChecker implements StatusChecker {

@Override
public Status check() {
// 获得服务集合
Collection<ExchangeServer> servers = DubboProtocol.getDubboProtocol().getServers();
// 如果为空则返回 UNKNOWN 的状态
if (servers == null || servers.isEmpty()) {
return new Status(Status.Level.UNKNOWN);
}
// 设置状态为 ok
Status.Level level = Status.Level.OK;
StringBuilder buf = new StringBuilder();
// 遍历集合
for (ExchangeServer server : servers) {
// 如果服务没有绑定到本地端口
if (!server.isBound()) {
// 状态改为 error
level = Status.Level.ERROR;
// 加入服务本地地址
buf.setLength(0);
buf.append(server.getLocalAddress());
break;
}
if (buf.length() > 0) {
buf.append(“,”);
}
// 如果服务绑定了本地端口,拼接 clients 数量
buf.append(server.getLocalAddress());
buf.append(“(clients:”);
buf.append(server.getChannels().size());
buf.append(“)”);
}
return new Status(level, buf.toString());
}

}

(十六)ThreadPoolStatusChecker
该类是对于线程池的状态进行监控。
@Activate
public class ThreadPoolStatusChecker implements StatusChecker {

@Override
public Status check() {
// 获得数据中心
DataStore dataStore = ExtensionLoader.getExtensionLoader(DataStore.class).getDefaultExtension();
// 获得线程池集合
Map<String, Object> executors = dataStore.get(Constants.EXECUTOR_SERVICE_COMPONENT_KEY);

StringBuilder msg = new StringBuilder();
// 设置为 ok
Status.Level level = Status.Level.OK;
// 遍历线程池集合
for (Map.Entry<String, Object> entry : executors.entrySet()) {
String port = entry.getKey();
ExecutorService executor = (ExecutorService) entry.getValue();

if (executor != null && executor instanceof ThreadPoolExecutor) {
ThreadPoolExecutor tp = (ThreadPoolExecutor) executor;
boolean ok = tp.getActiveCount() < tp.getMaximumPoolSize() – 1;
Status.Level lvl = Status.Level.OK;
// 如果活跃数量超过了最大的线程数量,则设置 warn
if (!ok) {
level = Status.Level.WARN;
lvl = Status.Level.WARN;
}

if (msg.length() > 0) {
msg.append(“;”);
}
// 输出线程池相关信息
msg.append(“Pool status:” + lvl
+ “, max:” + tp.getMaximumPoolSize()
+ “, core:” + tp.getCorePoolSize()
+ “, largest:” + tp.getLargestPoolSize()
+ “, active:” + tp.getActiveCount()
+ “, task:” + tp.getTaskCount()
+ “, service port: ” + port);
}
}
return msg.length() == 0 ? new Status(Status.Level.UNKNOWN) : new Status(level, msg.toString());
}

}

逻辑比较简单,我就不赘述了。
关于 telnet 下的相关实现请感兴趣的朋友直接查看,里面都是对于 telnet 命令的实现,内容比较独立。
后记
该部分相关的源码解析地址:https://github.com/CrazyHZM/i…

该文章讲解了远程调用中关于 dubbo 协议的部分,dubbo 协议是官方推荐使用的协议,并且对于 telnet 命令也做了很好的支持,要看懂这部分的逻辑,必须先对于之前的一些接口设计了解的很清楚。接下来我将开始对 rpc 模块的 dubbo-rpc-dubbo 关于 hessian 协议部分进行讲解。

正文完
 0