Thrift全面介绍 / 憋错料

官网：http://thrift.apache.org

简介

Thrift是一个软件框架，用来进行可扩展且跨语言的服务的开发。它结合了功能强大的软件堆栈和代码生成引擎，以构建在 C++, Java, Python, PHP, Ruby, Erlang, Perl, Haskell, C#, Cocoa, JavaScript, Node.js, Smalltalk, and OCaml 这些编程语言间无缝结合的、高效的服务。

Thrift最初由facebook开发用做系统内各语言之间的RPC通信。

2007年由facebook贡献到apache基金，08年5月进入apache孵化器。

支持多种语言之间的RPC方式的通信：php语言client可以构造一个对象，调用相应的服务方法来调用java语言的服务，跨越语言的C/S RPC调用。

Thrift允许定义一个简单的定义文件中的数据类型和服务接口，以作为输入文件，编译器生成代码用来方便地生成RPC客户端和服务器通信的无缝跨编程语言。

一个小实例

以UserService为例，描述一下使用thrift的方式，以及其原理。

service.thrift

struct User {
1:i64 id,
2:string name,
3:i64 timestamp,
4:bool vip
}
service UserService {
User getById(1:i64 id)
}

你可以将自己的Java服务通过".thrift"文件描述出来，并提供给服务消费端，那么消费端即可以生成自己的API文件。Thrift框架目前已经支持大部分主流的语言。需要注意，因为Thrift考虑到struct/service定义需要兼容多种语言的”风格"，所以它只支持一些基本的数据类型(比如i32,i64,string等)，以及service定义的方法不能重名，即使参数列表不同(并不是所有的语言都能像JAVA一样支持重载)。

生成API文件

首先下载和安装thrift客户端，比如在windows平台下，下载thrift.exe。不过此处需要提醒，不同的thrift客户端版本生成的API可能不兼容。本例使用thrift-0.9.0.exe，通过"--gen"指定生成API所适配的语言。本实例为生成java客户端API。

Java代码

//windows平台下,将API文件输出在service目录下(此目录需要存在)
> thrift.exe --gen java -o service service.thrift

需要明确的是：Thrift和其他RPC框架不同，thrift在生成的API文件中，已经描述了"调用过程"(即硬编码)，而不是像其他RPC那样在运行时(runtime)动态解析方法调用或者参数。

UserService实现类

Java代码

public class UserServiceImpl implements UserService.Iface {
@Override
public User getById(long id){
System.out.println("invoke...id:" + id);
return new User();//for test
}
}

实现类，需要放在Thrift server端。

原理简析

User.java: thrift生成API的能力还是非常的有限，比如在struct中只能使用简单的数据类型(不支持Date,Collection<?>等)，不过我们能从User中看出，它生成的类实现了"Serializable"接口和"TBase”接口。其中Serializable接口表明这个类的实例是需要序列化之后在网络中传输的，为了不干扰Java本身的序列化和反序列化机制，它还重写了readObject和writeObject方法，不过这对thrift本身并没有帮助。

TBase接口是thrift序列化和反序列化时使用的，它的两个核心方法：read和write。在上述的thrift文件中，struct定义的每个属性都有一个序号，比如1:id，那么thrift在序列化时，将会根据序号的顺序依次将属性的"名称 + 值"写入inputStream中,反序列化也是如此。(具体参见read和write的实现)。

Java代码

//read方法逐个读取字段,按照"索引",最终将"struct"对象封装完毕.
//write方法也非常类似,按照"索引"顺序逐个输出到流中.
while (true){
schemeField = iprot.readFieldBegin();
if (schemeField.type == org.apache.thrift.protocol.TType.STOP) {
break;
}
switch (schemeField.id) {
case 1: // ID
if (schemeField.type == org.apache.thrift.protocol.TType.I32) {
struct.id = iprot.readI32();
struct.setIdIsSet(true);
} else {
org.apache.thrift.protocol.TProtocolUtil.skip(iprot, schemeField.type);
}
break;
case 2: // NAME
..
}
}

因为thrift的序列化和反序列化实例数据时,是根据"属性序号"进行,这可以保证数据在inputstream和outputstream中顺序是严格的,此外每个struct中"序号"不能重复,但是可以不需要从"1"开始.如果"序号"有重复,将导致无法生成API文件.这一点也要求API开发者,如果更改了thrift文件中的struct定义,需要重新生成客户端API,否则服务将无法继续使用(可能报错,也可能数据错误).thrift序列化/反序列化的过程和JAVA自带的序列化机制不同,它将不会携带额外的class结构,此外thrift这种序列化机制更加适合网络传输,而且性能更加高效.

UserService.Client: 在生成的UserService中，有个Client静态类，这个类就是一个典型的代理类，此类已经实现了UserService的所有方法。开发者需要使用Client类中的API方法与Thrift server端交互，它将负责与Thrift server的Socket链接中，发送请求和接收响应。

需要注意的时，每次Client方法调用，都会在一个Socket链接中进行。这就意味着，在使用Client消费服务之前，需要和Thrift server建立有效的TCP链接。(稍后代码示例)

1) 发送请求:

Java代码

//参见:TServiceClient
//API方法调用时,发送请求数据流
protected void sendBase(String methodName, TBase args) throws TException {
oprot_.writeMessageBegin(new TMessage(methodName, TMessageType.CALL, ++seqid_));//首先写入"方法名称"和"seqid_"
args.write(oprot_);//序列化参数
oprot_.writeMessageEnd();
oprot_.getTransport().flush();
}
protected void receiveBase(TBase result, String methodName) throws TException {
TMessage msg = iprot_.readMessageBegin();//如果执行有异常
if (msg.type == TMessageType.EXCEPTION) {
TApplicationException x = TApplicationException.read(iprot_);
iprot_.readMessageEnd();
throw x;
}//检测seqid是否一致
if (msg.seqid != seqid_) {
throw new TApplicationException(TApplicationException.BAD_SEQUENCE_ID, methodName + " failed: out of sequence response");
}
result.read(iprot_);//反序列化
iprot_.readMessageEnd();
}

Thrift提供了简单的容错方式：每次方法调用，都会在Client端标记一个seqid，这是一个自增的本地ID，在TCP请求时将此seqid追加到流中，同时Server端响应时，也将此seqid原样返回过来；这样客户端就可以根据此值用来判断"请求--响应"是对应的，如果出现乱序，将会导致此请求以异常的方式结束。

2) 响应

Java代码

//参考: TBaseProcessor.java
@Override
public boolean process(TProtocol in, TProtocol out) throws TException {
TMessage msg = in.readMessageBegin();
ProcessFunction fn = processMap.get(msg.name);//根据方法名,查找"内部类"
if (fn == null) {
TProtocolUtil.skip(in, TType.STRUCT);
in.readMessageEnd();
TApplicationException x = new TApplicationException(TApplicationException.UNKNOWN_METHOD, "Invalid method name: ‘"+msg.name+"‘");
out.writeMessageBegin(new TMessage(msg.name, TMessageType.EXCEPTION, msg.seqid));
x.write(out);//序列化响应结果,直接输出
out.writeMessageEnd();
out.getTransport().flush();
return true;
}
fn.process(msg.seqid, in, out, iface);
return true;
}

thrift生成的UserService.Processor类,就是server端用来处理请求过程的"代理类";server端从socket中读取请求需要调用的"方法名" +参数列表,并交付给Processor类处理;和其他的RPC调用不同的时,thrift并没有使用类似于"反射机制"的方式来调用方法,而是将UserService的每个方法生成一个"内部类":

Java代码

public static class getById extends org.apache.thrift.ProcessFunction<I, getById_args> {
public getById() {
super("getById");//其中getById为标识符
}
public getById_args getEmptyArgsInstance() {
return new getById_args();
}
protected boolean isOneway() {
return false;
}
//实际处理方法
public getById_result getResult(I iface, getById_args args) throws org.apache.thrift.TException {
getById_result result = new getById_result();
result.success = iface.getById(args.id);
return result;
}
}

这个"内部类",将会在Processor初始化的时候,放入到一个map中,此后即可以通过"方法名"查找,然后调用其"getResult"方法了.

Java代码

public static class Processor extends org.apache.thrift.TBaseProcessor implements org.apache.thrift.TProcessor {
public Processor(I iface) {
super(iface, getProcessMap(new HashMap<String, org.apache.thrift.ProcessFunction<I, ? extends org.apache.thrift.TBase>>()));
}
protected Processor(I iface, Map<String, org.apache.thrift.ProcessFunction<I, ? extends org.apache.thrift.TBase>> processMap) {
super(iface, getProcessMap(processMap));
}
private static Map<String, org.apache.thrift.ProcessFunction<I, ? extends org.apache.thrift.TBase>> getProcessMap(Map<String, org.apache.thrift.ProcessFunction<I, ? extends org.apache.thrift.TBase>> processMap) {
//放入map
processMap.put("getById", new getById());
return processMap;
}
....
}

3) Server端Socket管理和执行策略

Java代码

TThreadPoolServer
public void serve() {
try {
//启动服务
serverTransport_.listen();
} catch (TTransportException ttx) {
LOGGER.error("Error occurred during listening.", ttx);
return;
}
// Run the preServe event
if (eventHandler_ != null) {
eventHandler_.preServe();
}
stopped_ = false;
setServing(true);
//循环,直到被关闭
while (!stopped_) {
int failureCount = 0;
try {
//accept客户端Socket链接,
//对于每个新链接,将会封装成runnable,并提交给线程或者线程池中运行.
TTransport client = serverTransport_.accept();
WorkerProcess wp = new WorkerProcess(client);
executorService_.execute(wp);
} catch (TTransportException ttx) {
if (!stopped_) {
++failureCount;
LOGGER.warn("Transport error occurred during acceptance of message.", ttx);
}
}
}
//....
}

Thrift Server端，设计思路也非常的直接。当前Service server启动之后，将会以阻塞的方式侦听Socket链接(代码参考TThreadPoolServer)，每建立一个Socket链接，都会将此Socket经过封装之后，放入线程池中，本质上也是一个Socket链接对应一个Worker Thread。这个Thread只会处理此Socket中的所有数据请求，直到Socket关闭。

Java代码

//参考:WorkerProcess
while (true) {
if (eventHandler != null) {
eventHandler.processContext(connectionContext, inputTransport, outputTransport);
}
if(stopped_ || !processor.process(inputProtocol, outputProtocol)) {
break;
}
}

当有Socket链接不是很多的时候，TThreadPoolServer并不会有太大的性能问题，可以通过指定ThreadPool中线程的个数进行简单的调优。如果Socket链接很多，我们只能使用TThreadedSelectorServer来做支撑，TThreadedSelectorServer内部基于NIO模式，具有异步的特性，可以极大的提升server端的并发能力；不过在绝大多数情况下，在thrift中使用"异步"似乎不太容易让人接受，毕竟这意味着Client端需要阻塞，并且在高并发环境中这个阻塞时间是不可控的。但SelecorServer确实可以有效的提升Server的并发能力，而且在一定程度上可以提升吞吐能力，这或许是我们优化Thrift Server比较可靠的方式之一。

Client端代码示例

Java代码

public class UserServiceClient {
public void startClient() {
TTransport transport;
try {
transport = new TSocket("localhost", 1234);
TProtocol protocol = new TBinaryProtocol(transport);
UserService.Client client = new UserService.Client(protocol);
transport.open();
User user = client.getById(1000);
////
transport.close();
} catch (TTransportException e) {
e.printStackTrace();
} catch (TException e) {
e.printStackTrace();
}
}
}

Server端代码示例

Java代码

public class Server {
public void startServer() {
try {
TServerSocket serverTransport = new TServerSocket(1234);
UserService.Processor process = new Processor(new UserServiceImpl());
Factory portFactory = new TBinaryProtocol.Factory(true, true);
Args args = new Args(serverTransport);
args.processor(process);
args.protocolFactory(portFactory);
TServer server = new TThreadPoolServer(args);
server.serve();
} catch (TTransportException e) {
e.printStackTrace();
}
}
}

到这里，你就会发现，一个service，需要server端启动一个ServerSocket，如果你有很多service，那么你需要让这些service尽可能的分布在不同的物理server上，否则一个物理server上运行太多的ServerSocket进程并不是一件让人愉快的事情.。或者你让几个service整合成一个。

问题总没有想象的那么简单，其实service被拆分的粒度越细，越容易被部署和扩展，对于负载均衡就更加有利。如何让一个service分布式部署，稍后再继续分享。

总结

1) Thrift文件定义struct和service API，此文件可以被其他语言生成API文件或者类文件。

2) 使用Thrift客户端生成API文件。

3) Java服务端(即服务提供端)，实现service功能。

4) 服务端将server发布成一个Thrift server：即将service嵌入到一个ServerSocket中。

5) 客户端启动Socket，并和Thrift server建立TCP连接，并使用Client代理类操作远程接口。

服务端开发

Thrift服务server端，其实就是一个ServerSocket线程+处理器，当Thrift-client端建立链接之后，处理器负责解析socket流信息，并根据其指定的"方法名"+参数列表，来调用"服务实现类”的方法，并将执行结果(或者异常)写入到socket中。

一个server，就需要创建一个ServerSocket，并侦听本地的一个端口，这种情况对分布式部署，有一些额外的要求：client端需要知道一个"service"被部署在了那些server上。

设计思路:

1) 每个server内部采用threadPool的方式,来提升并发能力.

2) 当server启动成功后,向zookeeper注册服务节点,此后client端就可以"感知到"服务的状态

3) 通过spring的方式,配置thrift-server服务类.

其中x注册是可选选项

1.pom.xml

<dependencies>
<dependency>
<groupId>org.springframework</groupId>
<artifactId>spring-context</artifactId>
<version>3.0.7.RELEASE</version>
</dependency>
<dependency>
<groupId>org.apache.zookeeper</groupId>
<artifactId>zookeeper</artifactId>
<version>3.4.5</version>
</dependency>
<!--
<dependency>
<groupId>com.101tec</groupId>
<artifactId>zkclient</artifactId>
<version>0.4</version>
</dependency>
-->
<dependency>
<groupId>org.apache.thrift</groupId>
<artifactId>libthrift</artifactId>
<version>0.9.1</version>
</dependency>
<dependency>
<groupId>org.apache.curator</groupId>
<artifactId>curator-recipes</artifactId>
<version>2.3.0</version>
</dependency>
<dependency>
<groupId>commons-pool</groupId>
<artifactId>commons-pool</artifactId>
<version>1.6</version>
</dependency>
</dependencies>

本实例,使用了apache-curator作为zookeeper客户端.

2. spring-thrift-server.xml

<bean id="thriftZookeeper" class="com.demo.thrift.zookeeper.ZookeeperFactory" destroy-method="close">
<property name="connectString" value="127.0.0.1:2181"></property>
<property name="namespace" value="demo/thrift-service"></property>
</bean>
<bean id="sericeAddressReporter" class="com.demo.thrift.support.impl.DynamicAddressReporter" destroy-method="close">
<property name="zookeeper" ref="thriftZookeeper"></property>
</bean>
<bean id="userService" class="com.demo.service.UserServiceImpl"/>
<bean class="com.demo.thrift.ThriftServiceServerFactory" destroy-method="close">
<property name="service" ref="userService"></property>
<property name="configPath" value="UserServiceImpl"></property>
<property name="port" value="9090"></property>
<property name="addressReporter" ref="sericeAddressReporter"></property>
</bean>

3. ThriftServiceServerFactory.java

此类严格上说并不是一个工厂类,它的主要作用就是封装指定的"service" ,然后启动一个server的过程,其中"service"属性表示服务的实现类,addressReporter表示当server启动成功后,需要指定的操作(比如,向zookeeper发送service的IP信息).

究竟当前server的ip地址是多少,在不同的设计中,有所不同,比如:有些管理员喜欢将本机的IP地址写入到os下的某个文件中,如果上层应用需要获取可靠的IP信息,就需要读取这个文件...你可以实现自己的ThriftServerIpTransfer来获取当前server的IP.

为了减少xml中的配置信息,在factory中,使用了反射机制来构建"Processor"类.

public class ThriftServiceServerFactory implements InitializingBean {
private Integer port;
private Integer priority = 1;// default
private Object service;// serice实现类
private ThriftServerIpTransfer ipTransfer;
private ThriftServerAddressReporter addressReporter;
private ServerThread serverThread;
private String configPath;
public void setService(Object service) {
this.service = service;
}
public void setPriority(Integer priority) {
this.priority = priority;
}
public void setPort(Integer port) {
this.port = port;
}
public void setIpTransfer(ThriftServerIpTransfer ipTransfer) {
this.ipTransfer = ipTransfer;
}
public void setAddressReporter(ThriftServerAddressReporter addressReporter) {
this.addressReporter = addressReporter;
}
public void setConfigPath(String configPath) {
this.configPath = configPath;
}
@Override
public void afterPropertiesSet() throws Exception {
if (ipTransfer == null) {
ipTransfer = new LocalNetworkIpTransfer();
}
String ip = ipTransfer.getIp();
if (ip == null) {
throw new NullPointerException("cant find server ip...");
}
String hostname = ip + ":" + port + ":" + priority;
Class serviceClass = service.getClass();
ClassLoader classLoader = Thread.currentThread().getContextClassLoader();
Class<?>[] interfaces = serviceClass.getInterfaces();
if (interfaces.length == 0) {
throw new IllegalClassFormatException("service-class should implements Iface");
}
// reflect,load "Processor";
Processor processor = null;
for (Class clazz : interfaces) {
String cname = clazz.getSimpleName();
if (!cname.equals("Iface")) {
continue;
}
String pname = clazz.getEnclosingClass().getName() + "$Processor";
try {
Class pclass = classLoader.loadClass(pname);
if (!pclass.isAssignableFrom(Processor.class)) {
continue;
}
Constructor constructor = pclass.getConstructor(clazz);
processor = (Processor) constructor.newInstance(service);
break;
} catch (Exception e) {
//
}
}
if (processor == null) {
throw new IllegalClassFormatException("service-class should implements Iface");
}
//需要单独的线程,因为serve方法是阻塞的.
serverThread = new ServerThread(processor, port);
serverThread.start();
// report
if (addressReporter != null) {
addressReporter.report(configPath, hostname);
}
}
class ServerThread extends Thread {
private TServer server;
ServerThread(Processor processor, int port) throws Exception {
TServerSocket serverTransport = new TServerSocket(port);
Factory portFactory = new TBinaryProtocol.Factory(true, true);
Args args = new Args(serverTransport);
args.processor(processor);
args.protocolFactory(portFactory);
server = new TThreadPoolServer(args);
}
@Override
public void run(){
try{
server.serve();
}catch(Exception e){
//
}
}
public void stopServer(){
server.stop();
}
}
public void close() {
serverThread.stopServer();
}
}

4. DynamicAddressReporter.java

在ThriftServiceServerFactory中,有个可选的属性:addressReporter, DynamicAddressReporter提供了向zookeeper注册service信息的能力,当server启动正常后,把server的IP + port发送到zookeeper中;那么此后服务消费client,就可以从zookeeper中获取server列表,并与它们建立链接(池).这样client端只需要关注zookeeper的节点名称即可,不需要配置大量的ip+port.

public class DynamicAddressReporter implements ThriftServerAddressReporter {
private CuratorFramework zookeeper;
public DynamicAddressReporter(){}
public DynamicAddressReporter(CuratorFramework zookeeper){
this.zookeeper = zookeeper;
}
public void setZookeeper(CuratorFramework zookeeper) {
this.zookeeper = zookeeper;
}
@Override
public void report(String service, String address) throws Exception {
if(zookeeper.getState() == CuratorFrameworkState.LATENT){
zookeeper.start();
zookeeper.newNamespaceAwareEnsurePath(service);
}
zookeeper.create()
.creatingParentsIfNeeded()
.withMode(CreateMode.EPHEMERAL_SEQUENTIAL)
.forPath(service +"/i_",address.getBytes("utf-8"));
}
public void close(){
zookeeper.close();
}
}

5. 测试类

public class ServiceMain {
/**
* @param args
*/
public static void main(String[] args) {
try {
ApplicationContext context = new ClassPathXmlApplicationContext("spring-thrift-server.xml");
Thread.sleep(3000000);
} catch (Exception e) {
e.printStackTrace();
}
}
}

客户端开发

Thrift-client作为服务消费端，由于thrift使用socket通讯，因此它需要面对几个问题：

1) client端需要知道server端的IP+port，如果是分布式部署，还需要知道所有server的IP+port列表。

2) client为了提升性能，不可能只使用一个socket来处理并发请求，当然也不能每个请求都创建一个socket；我们需要使用连接池方案。

3) 对于java开发工程师而言，基于spring配置thrift服务，可以提供很多的便利。

4) 基于zookeeper配置管理，那么client端就不需要"硬编码"的配置server的ip+port，可以使用zookeeper来推送每个service的服务地址。

5) 因为thrift-client端不使用连接池的话，将不能有效的提高并发能力，本文重点描述看如何使用thrift-client连接池。

1. pom.xml

<dependencies>
<dependency>
<groupId>org.springframework</groupId>
<artifactId>spring-context</artifactId>
<version>3.0.7.RELEASE</version>
</dependency>
<dependency>
<groupId>org.apache.zookeeper</groupId>
<artifactId>zookeeper</artifactId>
<version>3.4.5</version>
</dependency>
<!--
<dependency>
<groupId>com.101tec</groupId>
<artifactId>zkclient</artifactId>
<version>0.4</version>
</dependency>
-->
<dependency>
<groupId>org.apache.thrift</groupId>
<artifactId>libthrift</artifactId>
<version>0.9.1</version>
</dependency>
<dependency>
<groupId>org.apache.curator</groupId>
<artifactId>curator-recipes</artifactId>
<version>2.3.0</version>
</dependency>
<dependency>
<groupId>commons-pool</groupId>
<artifactId>commons-pool</artifactId>
<version>1.6</version>
</dependency>
</dependencies>

2. spring-thrift-client.xml

其中zookeeper作为可选项,开发者也可以通过制定serverAddress的方式指定server的地址.

<!--
<bean id="userService" class="com.demo.thrift.ThriftServiceClientProxyFactory">
<property name="service" value="com.demo.service.UserService"></property>
<property name="serverAddress" value="127.0.0.1:9090:2"></property>
<property name="maxActive" value="5"></property>
<property name="idleTime" value="10000"></property>
</bean>
-->
<bean id="thriftZookeeper" class="com.demo.thrift.zookeeper.ZookeeperFactory" destroy-method="close">
<property name="connectString" value="127.0.0.1:2181"></property>
<property name="namespace" value="demo/thrift-service"></property>
</bean>
<bean id="userService" class="com.demo.thrift.ThriftServiceClientProxyFactory" destroy-method="close">
<property name="service" value="com.demo.service.UserService"></property>
<property name="maxActive" value="5"></property>
<property name="idleTime" value="1800000"></property>
<property name="addressProvider">
<bean class="com.demo.thrift.support.impl.DynamicAddressProvider">
<property name="configPath" value="UserServiceImpl"></property>
<property name="zookeeper" ref="thriftZookeeper"></property>
</bean>
</property>
</bean>

3. ThriftServiceClientProxyFactory.java

因为我们要在client端使用连接池方案,那么就需要对client的方法调用过程,进行代理,这个类,就是维护了一个"Client"代理类,并在方法调用时,从"对象池"中取出一个"Client"对象,并在方法实际调用结束后归还给"对象池".

@SuppressWarnings("rawtypes")
public class ThriftServiceClientProxyFactory implements FactoryBean,InitializingBean {
private String service;
private String serverAddress;
private Integer maxActive = 32;//最大活跃连接数
////ms,default 3 min,链接空闲时间
//-1,关闭空闲检测
private Integer idleTime = 180000;
private ThriftServerAddressProvider addressProvider;
private Object proxyClient;
public void setMaxActive(Integer maxActive) {
this.maxActive = maxActive;
}
public void setIdleTime(Integer idleTime) {
this.idleTime = idleTime;
}
public void setService(String service) {
this.service = service;
}
public void setServerAddress(String serverAddress) {
this.serverAddress = serverAddress;
}
public void setAddressProvider(ThriftServerAddressProvider addressProvider) {
this.addressProvider = addressProvider;
}
private Class objectClass;
private GenericObjectPool<TServiceClient> pool;
private PoolOperationCallBack callback = new PoolOperationCallBack() {
@Override
public void make(TServiceClient client) {
System.out.println("create");
}
@Override
public void destroy(TServiceClient client) {
System.out.println("destroy");
}
};
@Override
public void afterPropertiesSet() throws Exception {
if(serverAddress != null){
addressProvider = new FixedAddressProvider(serverAddress);
}
ClassLoader classLoader = Thread.currentThread().getContextClassLoader();
//加载Iface接口
objectClass = classLoader.loadClass(service + "$Iface");
//加载Client.Factory类
Class<TServiceClientFactory<TServiceClient>> fi = (Class<TServiceClientFactory<TServiceClient>>)classLoader.loadClass(service + "$Client$Factory");
TServiceClientFactory<TServiceClient> clientFactory = fi.newInstance();
ThriftClientPoolFactory clientPool = new ThriftClientPoolFactory(addressProvider, clientFactory,callback);
GenericObjectPool.Config poolConfig = new GenericObjectPool.Config();
poolConfig.maxActive = maxActive;
poolConfig.minIdle = 0;
poolConfig.minEvictableIdleTimeMillis = idleTime;
poolConfig.timeBetweenEvictionRunsMillis = idleTime/2L;
pool = new GenericObjectPool<TServiceClient>(clientPool,poolConfig);
proxyClient = Proxy.newProxyInstance(classLoader,new Class[]{objectClass},new InvocationHandler() {
@Override
public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
//
TServiceClient client = pool.borrowObject();
try{
return method.invoke(client, args);
}catch(Exception e){
throw e;
}finally{
pool.returnObject(client);
}
}
});
}
@Override
public Object getObject() throws Exception {
return proxyClient;
}
@Override
public Class<?> getObjectType() {
return objectClass;
}
@Override
public boolean isSingleton() {
return true; //To change body of implemented methods use File | Settings | File Templates.
}
public void close(){
if(addressProvider != null){
addressProvider.close();
}
}
}

4. ThriftClientPoolFactory.java

"Client"对象池,对象池中是已经实例化的Client对象,Client对象负责与Thrift server通信.

/**
* 连接池,thrift-client for spring
*/
public class ThriftClientPoolFactory extends BasePoolableObjectFactory<TServiceClient>{
private final ThriftServerAddressProvider addressProvider;
private final TServiceClientFactory<TServiceClient> clientFactory;
private PoolOperationCallBack callback;
protected ThriftClientPoolFactory(ThriftServerAddressProvider addressProvider,TServiceClientFactory<TServiceClient> clientFactory) throws Exception {
this.addressProvider = addressProvider;
this.clientFactory = clientFactory;
}
protected ThriftClientPoolFactory(ThriftServerAddressProvider addressProvider,TServiceClientFactory<TServiceClient> clientFactory,PoolOperationCallBack callback) throws Exception {
this.addressProvider = addressProvider;
this.clientFactory = clientFactory;
this.callback = callback;
}
@Override
public TServiceClient makeObject() throws Exception {
InetSocketAddress address = addressProvider.selector();
TSocket tsocket = new TSocket(address.getHostName(),address.getPort());
TProtocol protocol = new TBinaryProtocol(tsocket);
TServiceClient client = this.clientFactory.getClient(protocol);
tsocket.open();
if(callback != null){
try{
callback.make(client);
}catch(Exception e){
//
}
}
return client;
}
public void destroyObject(TServiceClient client) throws Exception {
if(callback != null){
try{
callback.destroy(client);
}catch(Exception e){
//
}
}
TTransport pin = client.getInputProtocol().getTransport();
pin.close();
}
public boolean validateObject(TServiceClient client) {
TTransport pin = client.getInputProtocol().getTransport();
return pin.isOpen();
}
static interface PoolOperationCallBack {
//销毁client之前执行
void destroy(TServiceClient client);
//创建成功是执行
void make(TServiceClient client);
}
}

5. DynamicAddressProvider.java

将zookeeper作为server地址的提供者,这样客户端就不需要再配置文件中指定一堆ip + port,而且当server服务有更新时,也不需要client端重新配置.

/**
* 可以动态获取address地址,方案设计参考
* 1) 可以间歇性的调用一个web-service来获取地址
* 2) 可以使用事件监听机制,被动的接收消息,来获取最新的地址(比如基于MQ,nio等)
* 3) 可以基于zookeeper-watcher机制,获取最新地址
*
* 本实例,使用zookeeper作为"config"中心,使用apache-curator方法库来简化zookeeper开发
* 如下实现,仅供参考
*/
public class DynamicAddressProvider implements ThriftServerAddressProvider, InitializingBean {
private String configPath;
private PathChildrenCache cachedPath;
private CuratorFramework zookeeper;
//用来保存当前provider所接触过的地址记录
//当zookeeper集群故障时,可以使用trace中地址,作为"备份"
private Set<String> trace = new HashSet<String>();
private final List<InetSocketAddress> container = new ArrayList<InetSocketAddress>();
private Queue<InetSocketAddress> inner = new LinkedList<InetSocketAddress>();
private Object lock = new Object();
private static final Integer DEFAULT_PRIORITY = 1;
public void setConfigPath(String configPath) {
this.configPath = configPath;
}
public void setZookeeper(CuratorFramework zookeeper) {
this.zookeeper = zookeeper;
}
@Override
public void afterPropertiesSet() throws Exception {
//如果zk尚未启动,则启动
if(zookeeper.getState() == CuratorFrameworkState.LATENT){
zookeeper.start();
}
buildPathChildrenCache(zookeeper, configPath, true);
cachedPath.start(StartMode.POST_INITIALIZED_EVENT);
}
private void buildPathChildrenCache(CuratorFramework client, String path, Boolean cacheData) throws Exception {
cachedPath = new PathChildrenCache(client, path, cacheData);
cachedPath.getListenable().addListener(new PathChildrenCacheListener() {
@Override
public void childEvent(CuratorFramework client, PathChildrenCacheEvent event) throws Exception {
PathChildrenCacheEvent.Type eventType = event.getType();
switch (eventType) {
// case CONNECTION_RECONNECTED:
//
// break;
case CONNECTION_SUSPENDED:
case CONNECTION_LOST:
System.out.println("Connection error,waiting...");
return;
default:
//
}
//任何节点的时机数据变动,都会rebuild,此处为一个"简单的"做法.
cachedPath.rebuild();
rebuild();
}
protected void rebuild() throws Exception {
List<ChildData> children = cachedPath.getCurrentData();
if (children == null || children.isEmpty()) {
//有可能所有的thrift server都与zookeeper断开了链接
//但是,有可能,thrift client与thrift server之间的网络是良好的
//因此此处是否需要清空container,是需要多方面考虑的.
container.clear();
System.out.println("thrift server-cluster error....");
return;
}
List<InetSocketAddress> current = new ArrayList<InetSocketAddress>();
for (ChildData data : children) {
String address = new String(data.getData(), "utf-8");
current.addAll(transfer(address));
trace.add(address);
}
Collections.shuffle(current);
synchronized (lock) {
container.clear();
container.addAll(current);
inner.clear();
inner.addAll(current);
}
}
});
}
private List<InetSocketAddress> transfer(String address){
String[] hostname = address.split(":");
Integer priority = DEFAULT_PRIORITY;
if (hostname.length == 3) {
priority = Integer.valueOf(hostname[2]);
}
String ip = hostname[0];
Integer port = Integer.valueOf(hostname[1]);
List<InetSocketAddress> result = new ArrayList<InetSocketAddress>();
for (int i = 0; i < priority; i++) {
result.add(new InetSocketAddress(ip, port));
}
return result;
}
@Override
public List<InetSocketAddress> getAll() {
return Collections.unmodifiableList(container);
}
@Override
public synchronized InetSocketAddress selector() {
if (inner.isEmpty()) {
if(!container.isEmpty()){
inner.addAll(container);
}else if(!trace.isEmpty()){
synchronized (lock) {
for(String hostname : trace){
container.addAll(transfer(hostname));
}
Collections.shuffle(container);
inner.addAll(container);
}
}
}
return inner.poll();//null
}
@Override
public void close() {
try {
cachedPath.close();
zookeeper.close();
} catch (Exception e) {
//
}
}
}

到此为止，我们的Thrift基本上就可以顺利运行起来了。

使用ZooKeeper构建集群

通用办法是使用apache-curator组件来支持thrift连接zk提供集群服务并且推送服务信息变化。

其他高级话题

对于Thrift服务化的改造，主要是客户端，可以从如下几个方面进行：

1.服务端的服务注册，客户端自动发现，无需手工修改配置，这里我们使用zookeeper，但由于zookeeper本身提供的客户端使用较为复杂，因此采用curator-recipes工具类进行处理服务的注册与发现。

2.客户端使用连接池对服务调用进行管理，提升性能，这里我们使用Apache Commons项目commons-pool，可以大大减少代码的复杂度。

3.关于Failover/LoadBalance，由于zookeeper的watcher，当服务端不可用是及时通知客户端，并移除不可用的服务节点，而LoadBalance有很多算法，这里我们采用随机加权方式，也是常有的负载算法，至于其他的算法介绍参考:常见的负载均衡的基本算法。

4.使thrift服务的注册和发现可以基于spring配置,可以提供很多的便利。

5.其他的改造如：

1）通过动态代理实现client和server端的交互细节透明化，让用户只需通过服务方提供的接口进行访问

2）Thrift通过两种方式调用服务Client和Iface

// *) Client API 调用
(EchoService.Client)client.echo("hello lilei"); ---(1)
// *) Service 接口调用
(EchoService.Iface)service.echo("hello lilei"); ---(2)

Client API的方式, 不推荐, 我们推荐Service接口的方式(服务化)。

下面我们来一一实现：

一、pom.xml引入依赖jar包

<dependency>
<groupId>org.apache.thrift</groupId>
<artifactId>libthrift</artifactId>
<version>0.9.2</version>
</dependency>
<dependency>
<groupId>commons-pool</groupId>
<artifactId>commons-pool</artifactId>
<version>1.6</version>
</dependency>
<dependency>
<groupId>org.springframework</groupId>
<artifactId>spring-context</artifactId>
<version>4.0.9.RELEASE</version>
</dependency>
<dependency>
<groupId>org.apache.zookeeper</groupId>
<artifactId>zookeeper</artifactId>
<version>3.4.6</version>
</dependency>
<dependency>
<groupId>org.apache.curator</groupId>
<artifactId>curator-recipes</artifactId>
<version>2.7.1</version>
</dependency>

二、使用zookeeper管理服务节点配置

RPC服务往平台化的方向发展, 会屏蔽掉更多的服务细节(服务的IP地址集群, 集群的扩容和迁移), 只暴露服务接口. 这部分的演化, 使得server端和client端完全的解耦合. 两者的交互通过ConfigServer(MetaServer)的中介角色来搭线。

注: 该图源自dubbo的官网
这边借助Zookeeper来扮演该角色, server扮演发布者的角色, 而client扮演订阅者的角色.

Zookeeper是分布式应用协作服务. 它实现了paxos的一致性算法, 在命名管理/配置推送/数据同步/主从切换方面扮演重要的角色。其数据组织类似文件系统的目录结构:

每个节点被称为znode, 为znode节点依据其特性, 又可以分为如下类型:
　　1). PERSISTENT: 永久节点
　　2). EPHEMERAL: 临时节点, 会随session(client disconnect)的消失而消失
　　3). PERSISTENT_SEQUENTIAL: 永久节点, 其节点的名字编号是单调递增的
　　4). EPHEMERAL_SEQUENTIAL: 临时节点, 其节点的名字编号是单调递增的
　　注: 临时节点不能成为父节点
　　Watcher观察模式, client可以注册对节点的状态/内容变更的事件回调机制. 其Event事件的两类属性需要关注下:
　　1). KeeperState: Disconnected,SyncConnected,Expired
　　2). EventType: None,NodeCreated,NodeDeleted,NodeDataChanged,NodeChildrenChanged
RPC服务端:
　　作为具体业务服务的RPC服务发布方, 对其自身的服务描述由以下元素构成.
　　1). namespace: 命名空间，来区分不同应用
　　2). service: 服务接口, 采用发布方的类全名来表示
　　3). version: 版本号
　　借鉴了Maven的GAV坐标系, 三维坐标系更符合服务平台化的大环境.
　　*) 数据模型的设计
　　具体RPC服务的注册路径为: /rpc/{namespace}/{service}/{version}, 该路径上的节点都是永久节点
　　RPC服务集群节点的注册路径为: /rpc/{namespace}/{service}/{version}/{ip:port:weight}, 末尾的节点是临时节点.

1.定义Zookeeper的客户端的管理

ZookeeperFactory.java

package cn.slimsmart.thrift.rpc.zookeeper;

import org.apache.curator.framework.CuratorFramework;
import org.apache.curator.framework.CuratorFrameworkFactory;
import org.apache.curator.retry.ExponentialBackoffRetry;
import org.springframework.beans.factory.FactoryBean;
import org.springframework.util.StringUtils;
/**
* 获取zookeeper客户端链接
*/
public class ZookeeperFactory implements FactoryBean<CuratorFramework> {
private String zkHosts;
// session超时
private int sessionTimeout = 30000;
private int connectionTimeout = 30000;
// 共享一个zk链接
private boolean singleton = true;
// 全局path前缀,常用来区分不同的应用
private String namespace;
private final static String ROOT = "rpc";
private CuratorFramework zkClient;
public void setZkHosts(String zkHosts) {
this.zkHosts = zkHosts;
}
public void setSessionTimeout(int sessionTimeout) {
this.sessionTimeout = sessionTimeout;
}
public void setConnectionTimeout(int connectionTimeout) {
this.connectionTimeout = connectionTimeout;
}
public void setSingleton(boolean singleton) {
this.singleton = singleton;
}
public void setNamespace(String namespace) {
this.namespace = namespace;
}
public void setZkClient(CuratorFramework zkClient) {
this.zkClient = zkClient;
}
@Override
public CuratorFramework getObject() throws Exception {
if (singleton) {
if (zkClient == null) {
zkClient = create();
zkClient.start();
}
return zkClient;
}
return create();
}
@Override
public Class<?> getObjectType() {
return CuratorFramework.class;
}
@Override
public boolean isSingleton() {
return singleton;
}
public CuratorFramework create() throws Exception {
if (StringUtils.isEmpty(namespace)) {
namespace = ROOT;
} else {
namespace = ROOT +"/"+ namespace;
}
return create(zkHosts, sessionTimeout, connectionTimeout, namespace);
}
public static CuratorFramework create(String connectString, int sessionTimeout, int connectionTimeout, String namespace) {
CuratorFrameworkFactory.Builder builder = CuratorFrameworkFactory.builder();
return builder.connectString(connectString).sessionTimeoutMs(sessionTimeout).connectionTimeoutMs(30000)
.canBeReadOnly(true).namespace(namespace).retryPolicy(new ExponentialBackoffRetry(1000, Integer.MAX_VALUE))
.defaultData(null).build();
}
public void close() {
if (zkClient != null) {
zkClient.close();
}
}
}

2.服务端注册服务

由于服务端配置需要获取本机的IP地址，因此定义IP获取接口

ThriftServerIpResolve.java

package cn.slimsmart.thrift.rpc.zookeeper;
/**
*
* 解析thrift-server端IP地址，用于注册服务
* 1) 可以从一个物理机器或者虚机的特殊文件中解析
* 2) 可以获取指定网卡序号的Ip
* 3) 其他
*/
public interface ThriftServerIpResolve {
String getServerIp() throws Exception;
void reset();
//当IP变更时,将会调用reset方法
static interface IpRestCalllBack{
public void rest(String newIp);
}
}

可以对该接口做不通的实现，下面我们基于网卡获取IP地址，也可以通过配置serverIp
ThriftServerIpLocalNetworkResolve.java

package cn.slimsmart.thrift.rpc.zookeeper;
import java.net.Inet6Address;
import java.net.InetAddress;
import java.net.NetworkInterface;
import java.net.SocketException;
import java.util.Enumeration;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
/**
* 解析网卡Ip
*
*/
public class ThriftServerIpLocalNetworkResolve implements ThriftServerIpResolve {
private Logger logger = LoggerFactory.getLogger(getClass());
//缓存
private String serverIp;
public void setServerIp(String serverIp) {
this.serverIp = serverIp;
}
@Override
public String getServerIp() {
if (serverIp != null) {
return serverIp;
}
// 一个主机有多个网络接口
try {
Enumeration<NetworkInterface> netInterfaces = NetworkInterface.getNetworkInterfaces();
while (netInterfaces.hasMoreElements()) {
NetworkInterface netInterface = netInterfaces.nextElement();
// 每个网络接口,都会有多个"网络地址",比如一定会有lookback地址,会有siteLocal地址等.以及IPV4或者IPV6 .
Enumeration<InetAddress> addresses = netInterface.getInetAddresses();
while (addresses.hasMoreElements()) {
InetAddress address = addresses.nextElement();
if(address instanceof Inet6Address){
continue;
}
if (address.isSiteLocalAddress() && !address.isLoopbackAddress()) {
serverIp = address.getHostAddress();
logger.info("resolve server ip :"+ serverIp);
continue;
}
}
}
} catch (SocketException e) {
e.printStackTrace();
}
return serverIp;
}
@Override
public void reset() {
serverIp = null;
}
}

接下来我们定义发布服务接口，并实现将服务信息（服务接口、版本号，IP、port、weight）发布到zookeeper中。

ThriftServerAddressRegister.java

package cn.slimsmart.thrift.rpc.zookeeper;
/**
* 发布服务地址及端口到服务注册中心，这里是zookeeper服务器
*/
public interface ThriftServerAddressRegister {
/**
* 发布服务接口
* @param service 服务接口名称，一个产品中不能重复
* @param version 服务接口的版本号，默认1.0.0
* @param address 服务发布的地址和端口
*/
void register(String service,String version,String address);
}

实现：ThriftServerAddressRegisterZookeeper.java

package cn.slimsmart.thrift.rpc.zookeeper;
import java.io.UnsupportedEncodingException;
import org.apache.curator.framework.CuratorFramework;
import org.apache.curator.framework.imps.CuratorFrameworkState;
import org.apache.zookeeper.CreateMode;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.util.StringUtils;
import cn.slimsmart.thrift.rpc.ThriftException;
/**
* 注册服务列表到Zookeeper
*/
public class ThriftServerAddressRegisterZookeeper implements ThriftServerAddressRegister{
private Logger logger = LoggerFactory.getLogger(getClass());
private CuratorFramework zkClient;
public ThriftServerAddressRegisterZookeeper(){}
public ThriftServerAddressRegisterZookeeper(CuratorFramework zkClient){
this.zkClient = zkClient;
}
public void setZkClient(CuratorFramework zkClient) {
this.zkClient = zkClient;
}
@Override
public void register(String service, String version, String address) {
if(zkClient.getState() == CuratorFrameworkState.LATENT){
zkClient.start();
}
if(StringUtils.isEmpty(version)){
version="1.0.0";
}
//临时节点
try {
zkClient.create()
.creatingParentsIfNeeded()
.withMode(CreateMode.EPHEMERAL)
.forPath("/"+service+"/"+version+"/"+address);
} catch (UnsupportedEncodingException e) {
logger.error("register service address to zookeeper exception:{}",e);
throw new ThriftException("register service address to zookeeper exception: address UnsupportedEncodingException", e);
} catch (Exception e) {
logger.error("register service address to zookeeper exception:{}",e);
throw new ThriftException("register service address to zookeeper exception:{}", e);
}
}
public void close(){
zkClient.close();
}
}

3.客户端发现服务

定义获取服务地址接口

ThriftServerAddressProvider.java

package cn.slimsmart.thrift.rpc.zookeeper;
import java.net.InetSocketAddress;
import java.util.List;
/**
* thrift server-service地址提供者,以便构建客户端连接池
*/
public interface ThriftServerAddressProvider {
//获取服务名称
String getService();
/**
* 获取所有服务端地址
* @return
*/
List<InetSocketAddress> findServerAddressList();
/**
* 选取一个合适的address,可以随机获取等‘
* 内部可以使用合适的算法.
* @return
*/
InetSocketAddress selector();
void close();
}

基于zookeeper服务地址自动发现实现：ThriftServerAddressProviderZookeeper.java

package cn.slimsmart.thrift.rpc.zookeeper;
import java.net.InetSocketAddress;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashSet;
import java.util.LinkedList;
import java.util.List;
import java.util.Queue;
import java.util.Set;
import org.apache.curator.framework.CuratorFramework;
import org.apache.curator.framework.imps.CuratorFrameworkState;
import org.apache.curator.framework.recipes.cache.ChildData;
import org.apache.curator.framework.recipes.cache.PathChildrenCache;
import org.apache.curator.framework.recipes.cache.PathChildrenCache.StartMode;
import org.apache.curator.framework.recipes.cache.PathChildrenCacheEvent;
import org.apache.curator.framework.recipes.cache.PathChildrenCacheListener;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.InitializingBean;
/**
* 使用zookeeper作为"config"中心,使用apache-curator方法库来简化zookeeper开发
*/
public class ThriftServerAddressProviderZookeeper implements ThriftServerAddressProvider, InitializingBean {
private Logger logger = LoggerFactory.getLogger(getClass());
// 注册服务
private String service;
// 服务版本号
private String version = "1.0.0";
private PathChildrenCache cachedPath;
private CuratorFramework zkClient;
// 用来保存当前provider所接触过的地址记录
// 当zookeeper集群故障时,可以使用trace中地址,作为"备份"
private Set<String> trace = new HashSet<String>();
private final List<InetSocketAddress> container = new ArrayList<InetSocketAddress>();
private Queue<InetSocketAddress> inner = new LinkedList<InetSocketAddress>();
private Object lock = new Object();
// 默认权重
private static final Integer DEFAULT_WEIGHT = 1;
public void setService(String service) {
this.service = service;
}
public void setVersion(String version) {
this.version = version;
}
public ThriftServerAddressProviderZookeeper() {
}
public ThriftServerAddressProviderZookeeper(CuratorFramework zkClient) {
this.zkClient = zkClient;
}
public void setZkClient(CuratorFramework zkClient) {
this.zkClient = zkClient;
}
@Override
public void afterPropertiesSet() throws Exception {
// 如果zk尚未启动,则启动
if (zkClient.getState() == CuratorFrameworkState.LATENT) {
zkClient.start();
}
buildPathChildrenCache(zkClient, getServicePath(), true);
cachedPath.start(StartMode.POST_INITIALIZED_EVENT);
}
private String getServicePath(){
return "/" + service + "/" + version;
}
private void buildPathChildrenCache(final CuratorFramework client, String path, Boolean cacheData) throws Exception {
cachedPath = new PathChildrenCache(client, path, cacheData);
cachedPath.getListenable().addListener(new PathChildrenCacheListener() {
@Override
public void childEvent(CuratorFramework client, PathChildrenCacheEvent event) throws Exception {
PathChildrenCacheEvent.Type eventType = event.getType();
switch (eventType) {
case CONNECTION_RECONNECTED:
logger.info("Connection is reconection.");
break;
case CONNECTION_SUSPENDED:
logger.info("Connection is suspended.");
break;
case CONNECTION_LOST:
logger.warn("Connection error,waiting...");
return;
default:
//
}
// 任何节点的时机数据变动,都会rebuild,此处为一个"简单的"做法.
cachedPath.rebuild();
rebuild();
}
protected void rebuild() throws Exception {
List<ChildData> children = cachedPath.getCurrentData();
if (children == null || children.isEmpty()) {
// 有可能所有的thrift server都与zookeeper断开了链接
// 但是,有可能,thrift client与thrift server之间的网络是良好的
// 因此此处是否需要清空container,是需要多方面考虑的.
container.clear();
logger.error("thrift server-cluster error....");
return;
}
List<InetSocketAddress> current = new ArrayList<InetSocketAddress>();
String path = null;
for (ChildData data : children) {
path = data.getPath();
logger.debug("get path:"+path);
path = path.substring(getServicePath().length()+1);
logger.debug("get serviceAddress:"+path);
String address = new String(path.getBytes(), "utf-8");
current.addAll(transfer(address));
trace.add(address);
}
Collections.shuffle(current);
synchronized (lock) {
container.clear();
container.addAll(current);
inner.clear();
inner.addAll(current);
}
}
});
}
private List<InetSocketAddress> transfer(String address) {
String[] hostname = address.split(":");
Integer weight = DEFAULT_WEIGHT;
if (hostname.length == 3) {
weight = Integer.valueOf(hostname[2]);
}
String ip = hostname[0];
Integer port = Integer.valueOf(hostname[1]);
List<InetSocketAddress> result = new ArrayList<InetSocketAddress>();
// 根据优先级，将ip：port添加多次到地址集中，然后随机取地址实现负载
for (int i = 0; i < weight; i++) {
result.add(new InetSocketAddress(ip, port));
}
return result;
}
@Override
public List<InetSocketAddress> findServerAddressList() {
return Collections.unmodifiableList(container);
}
@Override
public synchronized InetSocketAddress selector() {
if (inner.isEmpty()) {
if (!container.isEmpty()) {
inner.addAll(container);
} else if (!trace.isEmpty()) {
synchronized (lock) {
for (String hostname : trace) {
container.addAll(transfer(hostname));
}
Collections.shuffle(container);
inner.addAll(container);
}
}
}
return inner.poll();
}
@Override
public void close() {
try {
cachedPath.close();
zkClient.close();
} catch (Exception e) {
}
}
@Override
public String getService() {
return service;
}
}

对此接口还做了一种实现，通过配置获取服务地址，参考附件：FixedAddressProvider.java

三、服务端服务注册实现

ThriftServiceServerFactory.java

package cn.slimsmart.thrift.rpc;
import java.lang.instrument.IllegalClassFormatException;
import java.lang.reflect.Constructor;
import org.apache.thrift.TProcessor;
import org.apache.thrift.TProcessorFactory;
import org.apache.thrift.protocol.TBinaryProtocol;
import org.apache.thrift.server.TServer;
import org.apache.thrift.server.TThreadedSelectorServer;
import org.apache.thrift.transport.TFramedTransport;
import org.apache.thrift.transport.TNonblockingServerSocket;
import org.springframework.beans.factory.InitializingBean;
import org.springframework.util.StringUtils;
import cn.slimsmart.thrift.rpc.zookeeper.ThriftServerAddressRegister;
import cn.slimsmart.thrift.rpc.zookeeper.ThriftServerIpLocalNetworkResolve;
import cn.slimsmart.thrift.rpc.zookeeper.ThriftServerIpResolve;
/**
* 服务端注册服务工厂
*/
public class ThriftServiceServerFactory implements InitializingBean {
// 服务注册本机端口
private Integer port = 8299;
// 优先级
private Integer weight = 1;// default
// 服务实现类
private Object service;// serice实现类
//服务版本号
private String version;
// 解析本机IP
private ThriftServerIpResolve thriftServerIpResolve;
//服务注册
private ThriftServerAddressRegister thriftServerAddressRegister;
private ServerThread serverThread;
public void setPort(Integer port) {
this.port = port;
}
public void setWeight(Integer weight) {
this.weight = weight;
}
public void setService(Object service) {
this.service = service;
}
public void setVersion(String version) {
this.version = version;
}
public void setThriftServerIpResolve(ThriftServerIpResolve thriftServerIpResolve) {
this.thriftServerIpResolve = thriftServerIpResolve;
}
public void setThriftServerAddressRegister(ThriftServerAddressRegister thriftServerAddressRegister) {
this.thriftServerAddressRegister = thriftServerAddressRegister;
}
@Override
public void afterPropertiesSet() throws Exception {
if (thriftServerIpResolve == null) {
thriftServerIpResolve = new ThriftServerIpLocalNetworkResolve();
}
String serverIP = thriftServerIpResolve.getServerIp();
if (StringUtils.isEmpty(serverIP)) {
throw new ThriftException("cant find server ip...");
}
String hostname = serverIP + ":" + port + ":" + weight;
Class<?> serviceClass = service.getClass();
// 获取实现类接口
Class<?>[] interfaces = serviceClass.getInterfaces();
if (interfaces.length == 0) {
throw new IllegalClassFormatException("service-class should implements Iface");
}
// reflect,load "Processor";
TProcessor processor = null;
String serviceName = null;
for (Class<?> clazz : interfaces) {
String cname = clazz.getSimpleName();
if (!cname.equals("Iface")) {
continue;
}
serviceName = clazz.getEnclosingClass().getName();
String pname = serviceName + "$Processor";
try {
ClassLoader classLoader = Thread.currentThread().getContextClassLoader();
Class<?> pclass = classLoader.loadClass(pname);
if (!TProcessor.class.isAssignableFrom(pclass)) {
continue;
}
Constructor<?> constructor = pclass.getConstructor(clazz);
processor = (TProcessor) constructor.newInstance(service);
break;
} catch (Exception e) {
//
}
}
if (processor == null) {
throw new IllegalClassFormatException("service-class should implements Iface");
}
//需要单独的线程,因为serve方法是阻塞的.
serverThread = new ServerThread(processor, port);
serverThread.start();
// 注册服务
if (thriftServerAddressRegister != null) {
thriftServerAddressRegister.register(serviceName, version, hostname);
}
}
class ServerThread extends Thread {
private TServer server;
ServerThread(TProcessor processor, int port) throws Exception {
TNonblockingServerSocket serverTransport = new TNonblockingServerSocket(port);
TThreadedSelectorServer.Args tArgs = new TThreadedSelectorServer.Args(serverTransport);
TProcessorFactory processorFactory = new TProcessorFactory(processor);
tArgs.processorFactory(processorFactory);
tArgs.transportFactory(new TFramedTransport.Factory());
tArgs.protocolFactory( new TBinaryProtocol.Factory(true, true));
server = new TThreadedSelectorServer(tArgs);
}
@Override
public void run(){
try{
//启动服务
server.serve();
}catch(Exception e){
//
}
}
public void stopServer(){
server.stop();
}
}
public void close() {
serverThread.stopServer();
}
}

四、客户端获取服务代理及连接池实现
客户端连接池实现：ThriftClientPoolFactory.java

package cn.slimsmart.thrift.rpc;
import java.net.InetSocketAddress;
import org.apache.commons.pool.BasePoolableObjectFactory;
import org.apache.thrift.TServiceClient;
import org.apache.thrift.TServiceClientFactory;
import org.apache.thrift.protocol.TBinaryProtocol;
import org.apache.thrift.protocol.TProtocol;
import org.apache.thrift.transport.TFramedTransport;
import org.apache.thrift.transport.TSocket;
import org.apache.thrift.transport.TTransport;
import cn.slimsmart.thrift.rpc.zookeeper.ThriftServerAddressProvider;
/**
* 连接池,thrift-client for spring
*/
public class ThriftClientPoolFactory extends BasePoolableObjectFactory<TServiceClient> {
private final ThriftServerAddressProvider serverAddressProvider;
private final TServiceClientFactory<TServiceClient> clientFactory;
private PoolOperationCallBack callback;
protected ThriftClientPoolFactory(ThriftServerAddressProvider addressProvider, TServiceClientFactory<TServiceClient> clientFactory) throws Exception {
this.serverAddressProvider = addressProvider;
this.clientFactory = clientFactory;
}
protected ThriftClientPoolFactory(ThriftServerAddressProvider addressProvider, TServiceClientFactory<TServiceClient> clientFactory,
PoolOperationCallBack callback) throws Exception {
this.serverAddressProvider = addressProvider;
this.clientFactory = clientFactory;
this.callback = callback;
}
static interface PoolOperationCallBack {
// 销毁client之前执行
void destroy(TServiceClient client);
// 创建成功是执行
void make(TServiceClient client);
}
public void destroyObject(TServiceClient client) throws Exception {
if (callback != null) {
try {
callback.destroy(client);
} catch (Exception e) {
//
}
}
TTransport pin = client.getInputProtocol().getTransport();
pin.close();
}
public boolean validateObject(TServiceClient client) {
TTransport pin = client.getInputProtocol().getTransport();
return pin.isOpen();
}
@Override
public TServiceClient makeObject() throws Exception {
InetSocketAddress address = serverAddressProvider.selector();
TSocket tsocket = new TSocket(address.getHostName(), address.getPort());
TTransport transport = new TFramedTransport(tsocket);
TProtocol protocol = new TBinaryProtocol(transport);
TServiceClient client = this.clientFactory.getClient(protocol);
transport.open();
if (callback != null) {
try {
callback.make(client);
} catch (Exception e) {
//
}
}
return client;
}
}

客户端服务代理工厂实现：ThriftServiceClientProxyFactory.java

package cn.slimsmart.thrift.rpc;
import java.lang.reflect.InvocationHandler;
import java.lang.reflect.Method;
import java.lang.reflect.Proxy;
import org.apache.commons.pool.impl.GenericObjectPool;
import org.apache.thrift.TServiceClient;
import org.apache.thrift.TServiceClientFactory;
import org.springframework.beans.factory.FactoryBean;
import org.springframework.beans.factory.InitializingBean;
import cn.slimsmart.thrift.rpc.ThriftClientPoolFactory.PoolOperationCallBack;
import cn.slimsmart.thrift.rpc.zookeeper.ThriftServerAddressProvider;
/**
* 客户端代理
*/
@SuppressWarnings({ "unchecked", "rawtypes" })
public class ThriftServiceClientProxyFactory implements FactoryBean, InitializingBean {
private Integer maxActive = 32;// 最大活跃连接数
// ms,default 3 min,链接空闲时间
// -1,关闭空闲检测
private Integer idleTime = 180000;
private ThriftServerAddressProvider serverAddressProvider;
private Object proxyClient;
private Class<?> objectClass;
private GenericObjectPool<TServiceClient> pool;
private PoolOperationCallBack callback = new PoolOperationCallBack() {
@Override
public void make(TServiceClient client) {
System.out.println("create");
}
@Override
public void destroy(TServiceClient client) {
System.out.println("destroy");
}
};
public void setMaxActive(Integer maxActive) {
this.maxActive = maxActive;
}
public void setIdleTime(Integer idleTime) {
this.idleTime = idleTime;
}
public void setServerAddressProvider(ThriftServerAddressProvider serverAddressProvider) {
this.serverAddressProvider = serverAddressProvider;
}
@Override
public void afterPropertiesSet() throws Exception {
ClassLoader classLoader = Thread.currentThread().getContextClassLoader();
// 加载Iface接口
objectClass = classLoader.loadClass(serverAddressProvider.getService() + "$Iface");
// 加载Client.Factory类
Class<TServiceClientFactory<TServiceClient>> fi = (Class<TServiceClientFactory<TServiceClient>>) classLoader.loadClass(serverAddressProvider.getService() + "$Client$Factory");
TServiceClientFactory<TServiceClient> clientFactory = fi.newInstance();
ThriftClientPoolFactory clientPool = new ThriftClientPoolFactory(serverAddressProvider, clientFactory, callback);
GenericObjectPool.Config poolConfig = new GenericObjectPool.Config();
poolConfig.maxActive = maxActive;
poolConfig.minIdle = 0;
poolConfig.minEvictableIdleTimeMillis = idleTime;
poolConfig.timeBetweenEvictionRunsMillis = idleTime / 2L;
pool = new GenericObjectPool<TServiceClient>(clientPool, poolConfig);
proxyClient = Proxy.newProxyInstance(classLoader, new Class[] { objectClass }, new InvocationHandler() {
@Override
public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
//
TServiceClient client = pool.borrowObject();
try {
return method.invoke(client, args);
} catch (Exception e) {
throw e;
} finally {
pool.returnObject(client);
}
}
});
}
@Override
public Object getObject() throws Exception {
return proxyClient;
}
@Override
public Class<?> getObjectType() {
return objectClass;
}
@Override
public boolean isSingleton() {
return true;
}
public void close() {
if (serverAddressProvider != null) {
serverAddressProvider.close();
}
}
}

下面我们看一下服务端和客户端的配置;

服务端spring-context-thrift-server.xml

<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:context="http://www.springframework.org/schema/context"
xmlns:aop="http://www.springframework.org/schema/aop" xmlns:tx="http://www.springframework.org/schema/tx"
xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-4.0.xsd
http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context-4.0.xsd
http://www.springframework.org/schema/aop http://www.springframework.org/schema/aop/spring-aop-4.0.xsd
http://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx-4.0.xsd"
default-lazy-init="false">
<bean id="thriftZookeeper" class="cn.slimsmart.thrift.rpc.zookeeper.ZookeeperFactory"
destroy-method="close">
<property name="zkHosts"
value="192.168.36.54:2181,192.168.36.99:2181,192.168.36.189:2181" />
<property name="namespace" value="cn.slimsmart.thrift.rpc.demo" />
<property name="connectionTimeout" value="3000" />
<property name="sessionTimeout" value="3000" />
<property name="singleton" value="true" />
</bean>
<bean id="sericeAddressRegister"
class="cn.slimsmart.thrift.rpc.zookeeper.ThriftServerAddressRegisterZookeeper"
destroy-method="close">
<property name="zkClient" ref="thriftZookeeper" />
</bean>
<bean id="echoSerivceImpl" class="cn.slimsmart.thrift.rpc.demo.EchoSerivceImpl" />
<bean id="echoSerivce" class="cn.slimsmart.thrift.rpc.ThriftServiceServerFactory"
destroy-method="close">
<property name="service" ref="echoSerivceImpl" />
<property name="port" value="9000" />
<property name="version" value="1.0.0" />
<property name="weight" value="1" />
<property name="thriftServerAddressRegister" ref="sericeAddressRegister" />
</bean>
<bean id="echoSerivce1" class="cn.slimsmart.thrift.rpc.ThriftServiceServerFactory"
destroy-method="close">
<property name="service" ref="echoSerivceImpl" />
<property name="port" value="9001" />
<property name="version" value="1.0.0" />
<property name="weight" value="1" />
<property name="thriftServerAddressRegister" ref="sericeAddressRegister" />
</bean>
<bean id="echoSerivce2" class="cn.slimsmart.thrift.rpc.ThriftServiceServerFactory"
destroy-method="close">
<property name="service" ref="echoSerivceImpl" />
<property name="port" value="9002" />
<property name="version" value="1.0.0" />
<property name="weight" value="1" />
<property name="thriftServerAddressRegister" ref="sericeAddressRegister" />
</bean>
</beans>

客户端：spring-context-thrift-client.xml

<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:context="http://www.springframework.org/schema/context"
xmlns:aop="http://www.springframework.org/schema/aop" xmlns:tx="http://www.springframework.org/schema/tx"
xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-4.0.xsd
http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context-4.0.xsd
http://www.springframework.org/schema/aop http://www.springframework.org/schema/aop/spring-aop-4.0.xsd
http://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx-4.0.xsd"
default-lazy-init="false">
<!--
<bean id="fixedAddressProvider" class="cn.slimsmart.thrift.rpc.zookeeper.FixedAddressProvider">
<property name="service" value="cn.slimsmart.thrift.rpc.demo.EchoSerivce" />
<property name="serverAddress" value="192.168.36.215:9001:1,192.168.36.215:9002:2,192.168.36.215:9003:3" />
</bean>
<bean id="echoSerivce" class="cn.slimsmart.thrift.rpc.ThriftServiceClientProxyFactory">
<property name="maxActive" value="5" />
<property name="idleTime" value="10000" />
<property name="serverAddressProvider" ref="fixedAddressProvider" />
</bean>
-->
<bean id="thriftZookeeper" class="cn.slimsmart.thrift.rpc.zookeeper.ZookeeperFactory"
destroy-method="close">
<property name="zkHosts"
value="192.168.36.54:2181,192.168.36.99:2181,192.168.36.189:2181" />
<property name="namespace" value="cn.slimsmart.thrift.rpc.demo" />
<property name="connectionTimeout" value="3000" />
<property name="sessionTimeout" value="3000" />
<property name="singleton" value="true" />
</bean>
<bean id="echoSerivce" class="cn.slimsmart.thrift.rpc.ThriftServiceClientProxyFactory" destroy-method="close">
<property name="maxActive" value="5" />
<property name="idleTime" value="1800000" />
<property name="serverAddressProvider">
<bean class="cn.slimsmart.thrift.rpc.zookeeper.ThriftServerAddressProviderZookeeper">
<property name="service" value="cn.slimsmart.thrift.rpc.demo.EchoSerivce" />
<property name="version" value="1.0.0" />
<property name="zkClient" ref="thriftZookeeper" />
</bean>
</property>
</bean>
</beans>