一、前言

前面介绍了Zookeeper的系统模型,下面进一步学习Zookeeper的底层序列化机制,Zookeeper的客户端与服务端之间会进行一系列的网络通信来实现数据传输,Zookeeper使用Jute组件来完成数据的序列化和反序列化操作。

二、Jute

Jute是Zookeeper底层序列化组件,其用于Zookeeper进行网络数据传输和本地磁盘数据存储的序列化和反序列化工作。

2.1 Jute序列化

MockReHeader实体类

复制代码
package com.hust.grid.leesf.jute.examples;

import org.apache.jute.InputArchive;
import org.apache.jute.OutputArchive;
import org.apache.jute.Record;

public class MockReHeader implements Record {
    private long sessionId;
    private String type;

    public MockReHeader() {

    }

    public MockReHeader(long sessionId, String type) {
        this.sessionId = sessionId;
        this.type = type;
    }

    public void setSessionId(long sessionId) {
        this.sessionId = sessionId;
    }

    public void setType(String type) {
        this.type = type;
    }

    public long getSessionId() {
        return sessionId;
    }

    public String getType() {
        return type;
    }

    public void serialize(OutputArchive oa, String tag) throws java.io.IOException {
        oa.startRecord(this, tag);
        oa.writeLong(sessionId, "sessionId");
        oa.writeString(type, "type");
        oa.endRecord(this, tag);
    }

    public void deserialize(InputArchive ia, String tag) throws java.io.IOException {
        ia.startRecord(tag);
        this.sessionId = ia.readLong("sessionId");
        this.type = ia.readString("type");
        ia.endRecord(tag);
    }

    @Override
    public String toString() {
        return "sessionId = " + sessionId + ", type = " + type;
    }
}
复制代码

Main

复制代码
package com.hust.grid.leesf.jute.examples;

import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.nio.ByteBuffer;

import org.apache.jute.BinaryInputArchive;
import org.apache.jute.BinaryOutputArchive;
import org.apache.zookeeper.server.ByteBufferInputStream;

public class Main {
    public static void main(String[] args) throws IOException {
        ByteArrayOutputStream baos = new ByteArrayOutputStream();
        BinaryOutputArchive boa = BinaryOutputArchive.getArchive(baos);
        new MockReHeader(0x3421eccb92a34el, "ping").serialize(boa, "header");

        ByteBuffer bb = ByteBuffer.wrap(baos.toByteArray());

        ByteBufferInputStream bbis = new ByteBufferInputStream(bb);
        BinaryInputArchive bia = BinaryInputArchive.getArchive(bbis);

        MockReHeader header2 = new MockReHeader();
        System.out.println(header2);
        header2.deserialize(bia, "header");
        System.out.println(header2);
        bbis.close();
        baos.close();
    }
}
复制代码

运行结果

sessionId = 0, type = null
sessionId = 14673999700337486, type = ping

说明:可以看到MockReHeader实体类需要实现Record接口并且实现serialize和deserialize方法。

OutputArchive和InputArchive分别是Jute底层的序列化器和反序列化器。

在Zookeeper的src文件夹下有zookeeper.jute文件,其内容如下

复制代码
/**
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

module org.apache.zookeeper.data {
    class Id {
        ustring scheme;
        ustring id;
    }
    class ACL {
        int perms;
        Id id;
    }
    // information shared with the client
    class Stat {
        long czxid;      // created zxid
        long mzxid;      // last modified zxid
        long ctime;      // created
        long mtime;      // last modified
        int version;     // version
        int cversion;    // child version
        int aversion;    // acl version
        long ephemeralOwner; // owner id if ephemeral, 0 otw
        int dataLength;  //length of the data in the node
        int numChildren; //number of children of this node
        long pzxid;      // last modified children
    }
    // information explicitly stored by the server persistently
    class StatPersisted {
        long czxid;      // created zxid
        long mzxid;      // last modified zxid
        long ctime;      // created
        long mtime;      // last modified
        int version;     // version
        int cversion;    // child version
        int aversion;    // acl version
        long ephemeralOwner; // owner id if ephemeral, 0 otw
        long pzxid;      // last modified children
    }

   // information explicitly stored by the version 1 database of servers
   class StatPersistedV1 {
       long czxid; //created zxid
       long mzxid; //last modified zxid
       long ctime; //created
       long mtime; //last modified
       int version; //version
       int cversion; //child version
       int aversion; //acl version
       long ephemeralOwner; //owner id if ephemeral. 0 otw
    }
}

module org.apache.zookeeper.proto {
    class ConnectRequest {
        int protocolVersion;
        long lastZxidSeen;
        int timeOut;
        long sessionId;
        buffer passwd;
    }
    class ConnectResponse {
        int protocolVersion;
        int timeOut;
        long sessionId;
        buffer passwd;
    }
    class SetWatches {
        long relativeZxid;
        vector<ustring>dataWatches;
        vector<ustring>existWatches;
        vector<ustring>childWatches;
    }
    class RequestHeader {
        int xid;
        int type;
    }
    class MultiHeader {
        int type;
        boolean done;
        int err;
    }
    class AuthPacket {
        int type;
        ustring scheme;
        buffer auth;
    }
    class ReplyHeader {
        int xid;
        long zxid;
        int err;
    }
    class GetDataRequest {
        ustring path;
        boolean watch;
    }
    class SetDataRequest {
        ustring path;
        buffer data;
        int version;
    }
    class SetDataResponse {
        org.apache.zookeeper.data.Stat stat;
    }
    class GetSASLRequest {
        buffer token;
    }
    class SetSASLRequest {
        buffer token;
    }
    class SetSASLResponse {
        buffer token;
    }
    class CreateRequest {
        ustring path;
        buffer data;
        vector<org.apache.zookeeper.data.ACL> acl;
        int flags;
    }
    class DeleteRequest {
        ustring path;
        int version;
    }
    class GetChildrenRequest {
        ustring path;
        boolean watch;
    }
    class GetChildren2Request {
        ustring path;
        boolean watch;
    }
    class CheckVersionRequest {
        ustring path;
        int version;
    }
    class GetMaxChildrenRequest {
        ustring path;
    }
    class GetMaxChildrenResponse {
        int max;
    }
    class SetMaxChildrenRequest {
        ustring path;
        int max;
    }
    class SyncRequest {
        ustring path;
    }
    class SyncResponse {
        ustring path;
    }
    class GetACLRequest {
        ustring path;
    }
    class SetACLRequest {
        ustring path;
        vector<org.apache.zookeeper.data.ACL> acl;
        int version;
    }
    class SetACLResponse {
        org.apache.zookeeper.data.Stat stat;
    }
    class WatcherEvent {
        int type;  // event type
        int state; // state of the Keeper client runtime
        ustring path;
    }
    class ErrorResponse {
        int err;
    }
    class CreateResponse {
        ustring path;
    }
    class ExistsRequest {
        ustring path;
        boolean watch;
    }
    class ExistsResponse {
        org.apache.zookeeper.data.Stat stat;
    }
    class GetDataResponse {
        buffer data;
        org.apache.zookeeper.data.Stat stat;
    }
    class GetChildrenResponse {
        vector<ustring> children;
    }
    class GetChildren2Response {
        vector<ustring> children;
        org.apache.zookeeper.data.Stat stat;
    }
    class GetACLResponse {
        vector<org.apache.zookeeper.data.ACL> acl;
        org.apache.zookeeper.data.Stat stat;
    }
}

module org.apache.zookeeper.server.quorum {
    class LearnerInfo {
        long serverid;
        int protocolVersion;
    }
    class QuorumPacket {
        int type; // Request, Ack, Commit, Ping
        long zxid;
        buffer data; // Only significant when type is request
        vector<org.apache.zookeeper.data.Id> authinfo;
    }
}

module org.apache.zookeeper.server.persistence {
    class FileHeader {
        int magic;
        int version;
        long dbid;
    }
}

module org.apache.zookeeper.txn {
    class TxnHeader {
        long clientId;
        int cxid;
        long zxid;
        long time;
        int type;
    }
    class CreateTxnV0 {
        ustring path;
        buffer data;
        vector<org.apache.zookeeper.data.ACL> acl;
        boolean ephemeral;
    }
    class CreateTxn {
        ustring path;
        buffer data;
        vector<org.apache.zookeeper.data.ACL> acl;
        boolean ephemeral;
        int parentCVersion;
    }
    class DeleteTxn {
        ustring path;
    }
    class SetDataTxn {
        ustring path;
        buffer data;
        int version;
    }
    class CheckVersionTxn {
        ustring path;
        int version;
    }
    class SetACLTxn {
        ustring path;
        vector<org.apache.zookeeper.data.ACL> acl;
        int version;
    }
    class SetMaxChildrenTxn {
        ustring path;
        int max;
    }
    class CreateSessionTxn {
        int timeOut;
    }
    class ErrorTxn {
        int err;
    }
    class Txn {
        int type;
        buffer data;
    }
    class MultiTxn {
        vector<org.apache.zookeeper.txn.Txn> txns;
    }
}
复制代码

其定义了所有的实体类的所属包名、类名及类的所有成员变量和类型,该文件会在源代码编译时,Jute会使用不同的代码生成器为这些类定义生成实际编程语言的类文件,如java语言生成的类文件保存在src/java/generated目录下,每个类都会实现Record接口。

三、通信协议

基于TCP/IP协议,Zookeeper实现了自己的通信协议来玩按成客户端与服务端、服务端与服务端之间的网络通信,对于请求,主要包含请求头和请求体,对于响应,主要包含响应头和响应体。

3.1 请求协议

对于请求协议而言,如下为获取节点数据请求的完整协议定义

    class RequestHeader {
        int xid;
        int type;
    }

从zookeeper.jute中可知RequestHeader包含了xid和type,xid用于记录客户端请求发起的先后序号,用来确保单个客户端请求的响应顺序,type代表请求的操作类型,如创建节点(OpCode.create)、删除节点(OpCode.delete)、获取节点数据(OpCode.getData)。

协议的请求主体内容部分,包含了请求的所有操作内容,不同的请求类型请求体不同。对于会话创建而言,其请求体如下

复制代码
    class ConnectRequest {
        int protocolVersion;
        long lastZxidSeen;
        int timeOut;
        long sessionId;
        buffer passwd;
    }
复制代码

Zookeeper客户端和服务器在创建会话时,会发送ConnectRequest请求,该请求包含协议版本号protocolVersion、最近一次接收到服务器ZXID lastZxidSeen、会话超时时间timeOut、会话标识sessionId和会话密码passwd。

对于获取节点数据而言,其请求体如下

    class GetDataRequest {
        ustring path;
        boolean watch;
    }

Zookeeper客户端在向服务器发送节点数据请求时,会发送GetDataRequest请求,该请求包含了数据节点路径path、是否注册Watcher的标识watch。

对于更新节点数据而言,其请求体如下

    class SetDataRequest {
        ustring path;
        buffer data;
        int version;
    }

Zookeeper客户端在向服务器发送更新节点数据请求时,会发送SetDataRequest请求,该请求包含了数据节点路径path、数据内容data、节点数据的期望版本号version。

针对不同的请求类型,Zookeeper都会定义不同的请求体,可以在zookeeper.jute中查看。

3.2 响应协议

对于响应协议而言,如下为获取节点数据响应的完整协议定义

响应头中包含了每个响应最基本的信息,包括xid、zxid和err:

    class ReplyHeader {
        int xid;
        long zxid;
        int err;
    }

xid与请求头中的xid一致,zxid表示Zookeeper服务器上当前最新的事务ID,err则是一个错误码,表示当请求处理过程出现异常情况时,就会在错误码中标识出来,常见的包括处理成功(Code.OK)、节点不存在(Code.NONODE)、没有权限(Code.NOAUTH)。

协议的响应主体内容部分,包含了响应的所有数据,不同的响应类型请求体不同。对于会话创建而言,其响应体如下

复制代码
    class ConnectResponse {
        int protocolVersion;
        int timeOut;
        long sessionId;
        buffer passwd;
    }
复制代码

针对客户端的会话创建请求,服务端会返回客户端一个ConnectResponse响应,该响应体包含了版本号protocolVersion、会话的超时时间timeOut、会话标识sessionId和会话密码passwd。

对于获取节点数据而言,其响应体如下

    class GetDataResponse {
        buffer data;
        org.apache.zookeeper.data.Stat stat;
    }

针对客户端的获取节点数据请求,服务端会返回客户端一个GetDataResponse响应,该响应体包含了数据节点内容data、节点状态stat。

对于更新节点数据而言,其响应体如下

    class SetDataResponse {
        org.apache.zookeeper.data.Stat stat;
    }

针对客户端的更新节点数据请求,服务端会返回客户端一个SetDataResponse响应,该响应体包含了最新的节点状态stat。

针对不同的响应类型,Zookeeper都会定义不同的响应体,可以在zookeeper.jute中查看。

四、总结

本篇博客讲解了Zookeeper中的序列化机制和客户端与服务端、服务端与服务端的通信协议,内容相对较为简单,容易理解,谢谢各位园友的观看~

One thought on “【分布式】Zookeeper序列化及通信协议”

Leave a Reply

Your email address will not be published. Required fields are marked *