1. 程式人生 > >布式架構基礎:Java RMI詳解

布式架構基礎:Java RMI詳解

RMI簡介 ​ Java RMI,即 遠端方法呼叫(Remote Method Invocation),一種用於實現遠端過程呼叫(RPC)(Remote procedure call)的Java API, 能直接傳輸序列化後的Java物件和分散式垃圾收集。它的實現依賴於Java虛擬機器(JVM),因此它僅支援從一個JVM到另一個JVM的呼叫。

這裡寫圖片描述 rmi的實現 (1) 直接使用Registry實現rmi 服務端:

public class RegistryService { public static void main(String[] args) { try { // 本地主機上的遠端物件登錄檔Registry的例項,預設埠1099 Registry registry = LocateRegistry.createRegistry(1099); // 建立一個遠端物件 HelloRegistryFacade hello = new HelloRegistryFacadeImpl(); // 把遠端物件註冊到RMI註冊伺服器上,並命名為HelloRegistry registry.rebind(“HelloRegistry”, hello); System.out.println(“======= 啟動RMI服務成功! =======”); } catch (RemoteException e) { e.printStackTrace(); } } } 介面: 繼承Remote介面 public interface HelloRegistryFacade extends Remote {

String helloWorld(String name) throws RemoteException;

} 介面實現: 繼承UnicastRemoteObject public class HelloRegistryFacadeImpl extends UnicastRemoteObject implements HelloRegistryFacade{

public HelloRegistryFacadeImpl() throws RemoteException {
    super();
}

@Override
public String helloWorld(String name) {
    return "[Registry] 你好! " + name;
}

} 客戶端: public class RegistryClient { public static void main(String[] args) { try { Registry registry = LocateRegistry.getRegistry(1099); HelloRegistryFacade hello = (HelloRegistryFacade) registry.lookup(“HelloRegistry”); String response = hello.helloWorld(“ZhenJin”); System.out.println(“=======> ” + response + ” <=======”); } catch (NotBoundException | RemoteException e) { e.printStackTrace(); } } }

圖解: 這裡寫圖片描述 Registry(登錄檔)是放置所有伺服器物件的名稱空間。 每次服務端建立一個物件時,它都會使用bind()或rebind()方法註冊該物件。 這些是使用稱為繫結名稱的唯一名稱註冊的。

要呼叫遠端物件,客戶端需要該物件的引用,如(HelloRegistryFacade)。 即通過服務端繫結的名稱(HelloRegistry)從登錄檔中獲取物件(lookup()方法)。 (2) 使用Naming方法實現rmi 服務端:

public class NamingService { public static void main(String[] args) { try { // 本地主機上的遠端物件登錄檔Registry的例項 LocateRegistry.createRegistry(1100); // 建立一個遠端物件 HelloNamingFacade hello = new HelloNamingFacadeImpl(); // 把遠端物件註冊到RMI註冊伺服器上,並命名為Hello //繫結的URL標準格式為:rmi://host:port/name Naming.bind(“rmi://localhost:1100/HelloNaming”, hello); System.out.println(“======= 啟動RMI服務成功! =======”); } catch (RemoteException | MalformedURLException | AlreadyBoundException e) { e.printStackTrace(); } } } 介面和介面實現和Registry的方式一樣

客戶端:

public class NamingClient { public static void main(String[] args) { try { String remoteAddr=”rmi://localhost:1100/HelloNaming”; HelloNamingFacade hello = (HelloNamingFacade) Naming.lookup(remoteAddr); String response = hello.helloWorld(“ZhenJin”); System.out.println(“=======> ” + response + ” <=======”); } catch (NotBoundException | RemoteException | MalformedURLException e) { e.printStackTrace(); } } } Naming部分原始碼: public static Remote lookup(String name) throws NotBoundException,java.net.MalformedURLException,RemoteException{ ParsedNamingURL parsed = parseURL(name); Registry registry = getRegistry(parsed);

if (parsed.name == null)
    return registry;
return registry.lookup(parsed.name); }

Naming其實是對Registry的一個封裝

Scala實現rmi 上面說了rmi是通過JVM虛擬機器進行一個遠端呼叫的,我們通過Scala,kotlin等jvm語言印證下

服務端:

object ScalaRmiService extends App { try { val user:UserScalaFacade = new UserScalaFacadeImpl LocateRegistry.createRegistry(1103) Naming.rebind(“rmi://localhost:1103/UserScala”, user) println(“======= 啟動RMI服務成功! =======”) } catch { case e: IOException => println(e) } } 介面 trait UserScalaFacade extends Remote {

/** * 通過使用者名稱獲取使用者資訊 */ @throws(classOf[RemoteException]) def getByName(userName: String): User

/** * 通過使用者性別獲取使用者資訊 */ @throws(classOf[RemoteException]) def getBySex(userSex: String): List[User]

} 介面實現: class UserScalaFacadeImpl extends UnicastRemoteObject with UserScalaFacade {

/** * 模擬一個數據庫表 */ private lazy val userList = List( new User(“Jane”, “女”, 16), new User(“jack”, “男”, 17), new User(“ZhenJin”, “男”, 18) )

override def getByName(userName: String): User = userList.filter(u => userName.equals(u.userName)).head

override def getBySex(userSex: String): List[User] = userList.filter(u => userSex.equals(u.userSex))

} 實體類: 實體類必須實現序列化(Serializable)才能進行一個遠端傳輸 class User(name: String, sex: String, age: Int) extends Serializable {

var userName: String = name var userSex: String = sex var userAge: Int = age override def toString = s”User(userName=userName,userSex=userSex, userAge=$userAge)”

} Scala客戶端: object ScalaRmiClient extends App {

try {

val remoteAddr="rmi://localhost:1103/UserScala"
val userFacade = Naming.lookup(remoteAddr).asInstanceOf[UserScalaFacade]

println(userFacade.getByName("ZhenJin"))
System.out.println("--------------------------------------")
for (user <- userFacade.getBySex("男")) println(user)

} catch { case e: NotBoundException => println(e) case e: RemoteException => println(e) case e: MalformedURLException => println(e) }

} Java客戶端: public class JavaRmiClient {

public static void main(String[] args) {

    try {
        String remoteAddr="rmi://localhost:1103/UserScala";
        UserScalaFacade userFacade = (UserScalaFacade) Naming.lookup();

        User zhenJin = userFacade.getByName("ZhenJin");
        System.out.println(zhenJin);
        System.out.println("--------------------------------------");
        List<User> userList = userFacade.getBySex("男");
        System.out.println(userList);

    } catch (NotBoundException | RemoteException | MalformedURLException e) {
        e.printStackTrace();
    }
} }

上面試驗可以證明Scala和Java是可以互通的,Scala本身也是可以直接引用Java類的

序列化簡介 ​ 序列化(Serialization)是將資料結構或物件狀態轉換為可以儲存(例如,在檔案或儲存器緩衝區中)或傳輸(例如,通過網路連線)的格式的過程, 反序列化(Deserialization)則是從一系列位元組中提取資料結構的相反操作.

這裡寫圖片描述 Kotlin實現rmi 服務端:

fun main(args: Array) { try { val hello: HelloKotlinFacade = HelloKotlinFacadeImpl() LocateRegistry.createRegistry(1102) Naming.rebind(“rmi://localhost:1101/HelloKotlin”, hello) println(“======= 啟動RMI服務成功! =======”) } catch (e: IOException) { e.printStackTrace() } } 客戶端: fun main(args: Array) { try { val hello = Naming.lookup(“rmi://localhost:1102/HelloKotlin”) as HelloKotlinFacade val response = hello.helloWorld(“ZhenJin”) println(“=======> $response <=======”) } catch (e: NotBoundException) { e.printStackTrace() } catch (e: RemoteException) { e.printStackTrace() } catch (e: MalformedURLException) { e.printStackTrace() } } 實現和介面省略…

SpringBoot實現rmi StringBoot通過配置就可以簡單實現rmi了

服務端:

@Configuration public class RmiServiceConfig { @Bean public RmiServiceExporter registerService(UserFacade userFacade) { RmiServiceExporter rmiServiceExporter = new RmiServiceExporter(); rmiServiceExporter.setServiceName(“UserInfo”); rmiServiceExporter.setService(userFacade); rmiServiceExporter.setServiceInterface(UserFacade.class); rmiServiceExporter.setRegistryPort(1101); return rmiServiceExporter; } } 客戶端: @Configuration public class RmiClientConfig {

@Bean
public UserFacade userInfo() {
    RmiProxyFactoryBean rmiProxyFactoryBean = new RmiProxyFactoryBean();
    rmiProxyFactoryBean.setServiceUrl("rmi://localhost:1101/UserInfo");
    rmiProxyFactoryBean.setServiceInterface(UserFacade.class);
    rmiProxyFactoryBean.afterPropertiesSet();
    return (UserFacade) rmiProxyFactoryBean.getObject();
}

} 客戶端測試類: @Autowired private UserFacade userFacade; @Test public void userBySexTest() { try { List userList = userFacade.getBySex(“男”); userList.forEach(System.out::println); } catch (RemoteException e) { e.printStackTrace(); } } 通過測試類可以看出,這和我們平時的程式呼叫內部方法沒什麼區別!

rmi呼叫過程 這裡寫圖片描述 有兩個遠端服務介面可供client呼叫,Factory和Product介面

FactoryImpl類實現了Factory介面,ProductImpl類實現了Product介面 1. FactoryImpl被註冊到了rmi-registry中 2. client端請求一個Factory的引用 3. rmi-registry返回client端一個FactoryImpl的引用 4. client端呼叫FactoryImpl的遠端方法請求一個ProductImpl的遠端引用 5. FactoryImpl返回給client端一個ProductImpl引用 6. client通過ProductImpl引用呼叫遠端方法 Zookeeper實現rmi 出處:http://www.importnew.com/20344.html

安裝Zookeeper 解壓 ZooKeeper

tar -zxvf zookeeper-3.4.12.tar.gz

在 conf 目錄新建 zoo.cfg

cd zookeeper-3.4.12/conf vim zoo.cfg

zoo.cfg 程式碼如下(自己指定 log 檔案目錄):

tickTime=2000 dataDir=/usr/local/zookeeper-3.4.12/data dataLogDir=/usr/local/zookeeper-3.4.12/log clientPort=2181

在 bin 目錄下,啟動 Zookeeper:

cd zookeeper-3.4.12/bin ./zkServer.sh start

消費者:

public class RmiConsumer {

// 用於等待 SyncConnected 事件觸發後繼續執行當前執行緒
private CountDownLatch latch = new CountDownLatch(1);

// 定義一個 volatile 成員變數,用於儲存最新的 RMI 地址(考慮到該變數或許會被其它執行緒所修改,一旦修改後,該變數的值會影響到所有執行緒)
private volatile List<String> urlList = new ArrayList<>();

// 構造器
public RmiConsumer() {
    ZooKeeper zk = connectServer(); // 連線 ZooKeeper 伺服器並獲取 ZooKeeper 物件
    if (zk != null) {
        watchNode(zk); // 觀察 /registry 節點的所有子節點並更新 urlList 成員變數
    }
}

// 查詢 RMI 服務
public <T extends Remote> T lookup() {
    T service = null;
    int size = urlList.size();
    if (size > 0) {
        String url;
        if (size == 1) {
            url = urlList.get(0); // 若 urlList 中只有一個元素,則直接獲取該元素
            log.debug("using only url: {}", url);
        } else {
            url = urlList.get(ThreadLocalRandom.current().nextInt(size)); // 若 urlList

中存在多個元素,則隨機獲取一個元素 log.debug(“using random url: {}”, url); } service = lookupService(url); // 從 JNDI 中查詢 RMI 服務 } return service; }

// 連線 ZooKeeper 伺服器
private ZooKeeper connectServer() {
    ZooKeeper zk = null;
    try {
        zk = new ZooKeeper(Constant.ZK_CONNECTION_STRING, Constant.ZK_SESSION_TIMEOUT, new Watcher() {
            @Override
            public void process(WatchedEvent event) {
                if (event.getState() == Event.KeeperState.SyncConnected) {
                    latch.countDown(); // 喚醒當前正在執行的執行緒
                }
            }
        });
        latch.await(); // 使當前執行緒處於等待狀態
    } catch (IOException | InterruptedException e) {
        log.error("", e);
    }
    return zk;
}

// 觀察 /registry 節點下所有子節點是否有變化
private void watchNode(final ZooKeeper zk) {
    try {
        List<String> nodeList = zk.getChildren(Constant.ZK_REGISTRY_PATH, event -> {
            if (event.getType() == Watcher.Event.EventType.NodeChildrenChanged) {
                watchNode(zk); // 若子節點有變化,則重新呼叫該方法(為了獲取最新子節點中的資料)
            }
        });
        List<String> dataList = new ArrayList<>(); // 用於存放 /registry 所有子節點中的資料
        for (String node : nodeList) {
            byte[] data = zk.getData(Constant.ZK_REGISTRY_PATH + "/" + node, false, null); // 獲取 /registry 的子節點中的資料
            dataList.add(new String(data));
        }
        log.debug("node data: {}", dataList);
        urlList = dataList; // 更新最新的 RMI 地址
    } catch (KeeperException | InterruptedException e) {
        log.error("", e);
    }
}

// 在 JNDI 中查詢 RMI 遠端服務物件
@SuppressWarnings("unchecked")
private <T> T lookupService(String url) {
    T remote = null;
    try {
        remote = (T) Naming.lookup(url);
    } catch (NotBoundException | MalformedURLException | RemoteException e) {
        log.error("遠端查找出錯!", e);
    }
    return remote;
} }

生產者: public class RmiProvider {

/**
 * 用於等待 SyncConnected 事件觸發後繼續執行當前執行緒
 */
private CountDownLatch latch = new CountDownLatch(1);

// 釋出 RMI 服務並註冊 RMI 地址到 ZooKeeper 中
public void publish(Remote remote, String host, int port) {
    String url = publishService(remote, host, port); // 釋出 RMI 服務並返回 RMI 地址
    if (url != null) {
        ZooKeeper zk = connectServer(); // 連線 ZooKeeper 伺服器並獲取 ZooKeeper 物件
        if (zk != null) {
            createNode(zk, url); // 建立 ZNode 並將 RMI 地址放入 ZNode 上
        }
    }
}

 /**
  *釋出 RMI 服務
  */
private String publishService(Remote remote, String host, int port) {
    String url = null;
    try {
        url = String.format("rmi://%s:%d/%s", host, port, remote.getClass().getName());
        LocateRegistry.createRegistry(port);
        Naming.rebind(url, remote);
        log.debug("publish rmi service (url: {})", url);
    } catch (RemoteException | MalformedURLException e) {
        log.error("", e);
    }
    return url;
}

// 連線 ZooKeeper 伺服器
private ZooKeeper connectServer() {
    ZooKeeper zk = null;
    try {
        zk = new ZooKeeper(Constant.ZK_CONNECTION_STRING, Constant.ZK_SESSION_TIMEOUT, new Watcher() {
            @Override
            public void process(WatchedEvent event) {
                if (event.getState() == Event.KeeperState.SyncConnected) {
                    latch.countDown(); // 喚醒當前正在執行的執行緒
                }
            }
        });
        latch.await(); // 使當前執行緒處於等待狀態
    } catch (IOException | InterruptedException e) {
        log.error("", e);
    }
    return zk;
}

/**
 * 建立節點
 */
private void createNode(ZooKeeper zk, String url) {
    try {
        byte[] data = url.getBytes();
        String path = zk.create(Constant.ZK_PROVIDER_PATH, data, ZooDefs.Ids.OPEN_ACL_UNSAFE, CreateMode.EPHEMERAL_SEQUENTIAL);    // 建立一個臨時性且有序的 ZNode
        log.debug("create zookeeper node ({} => {})", path, url);
    } catch (KeeperException | InterruptedException e) {
        log.error("", e);
    }
}

} 圖解: 這裡寫圖片描述