Skip to content

Raft Java implementation which is simple and easy to understand.

License

Notifications You must be signed in to change notification settings

jzdayz/raft-java

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

raft-java

Raft implementation library for Java.
参考自Raft论文和Raft作者的开源实现LogCabin

支持的功能

  • leader选举
  • 日志复制
  • snapshot
  • 集群成员动态更变

Quick Start

在本地单机上部署一套3实例的raft集群,执行如下脚本:
cd raft-java-example && sh deploy.sh
该脚本会在raft-java-example/env目录部署三个实例example1、example2、example3;
同时会创建一个client目录,用于测试raft集群读写功能。
部署成功后,测试写操作,通过如下脚本: cd env/client
./bin/run_client.sh "list://127.0.0.1:8051,127.0.0.1:8052,127.0.0.1:8053" hello world
测试读操作命令:
./bin/run_client.sh "list://127.0.0.1:8051,127.0.0.1:8052,127.0.0.1:8053" hello

使用方法

下面介绍如何在代码中使用raft-java依赖库来实现一套分布式存储系统。

配置依赖(暂未发不到maven中央仓库,需要手动install到本地)

<dependency>
    <groupId>com.github.wenweihu86.raft</groupId>
    <artifactId>raft-java-core</artifactId>
    <version>1.9.0</version>
</dependency>

定义数据写入和读取接口

message SetRequest {
    string key = 1;
    string value = 2;
}
message SetResponse {
    bool success = 1;
}
message GetRequest {
    string key = 1;
}
message GetResponse {
    string value = 1;
}
public interface ExampleService {
    Example.SetResponse set(Example.SetRequest request);
    Example.GetResponse get(Example.GetRequest request);
}

服务端使用方法

  1. 实现状态机StateMachine接口实现类
// 该接口三个方法主要是给Raft内部调用
public interface StateMachine {
    /**
     * 对状态机中数据进行snapshot,每个节点本地定时调用
     * @param snapshotDir snapshot数据输出目录
     */
    void writeSnapshot(String snapshotDir);
    /**
     * 读取snapshot到状态机,节点启动时调用
     * @param snapshotDir snapshot数据目录
     */
    void readSnapshot(String snapshotDir);
    /**
     * 将数据应用到状态机
     * @param dataBytes 数据二进制
     */
    void apply(byte[] dataBytes);
}
  1. 实现数据写入和读取接口
// ExampleService实现类中需要包含以下成员
private RaftNode raftNode;
private ExampleStateMachine stateMachine;
// 数据写入主要逻辑
byte[] data = request.toByteArray();
// 数据同步写入raft集群
boolean success = raftNode.replicate(data, Raft.EntryType.ENTRY_TYPE_DATA);
Example.SetResponse response = Example.SetResponse.newBuilder().setSuccess(success).build();
// 数据读取主要逻辑,由具体应用状态机实现
Example.GetResponse response = stateMachine.get(request);
  1. 服务端启动逻辑
// 初始化RPCServer
RPCServer server = new RPCServer(localServer.getEndPoint().getPort());
// 应用状态机
ExampleStateMachine stateMachine = new ExampleStateMachine();
// 设置Raft选项,比如:
RaftOptions.snapshotMinLogSize = 10 * 1024;
RaftOptions.snapshotPeriodSeconds = 30;
RaftOptions.maxSegmentFileSize = 1024 * 1024;
// 初始化RaftNode
RaftNode raftNode = new RaftNode(serverList, localServer, stateMachine);
// 注册Raft节点之间相互调用的服务
RaftConsensusService raftConsensusService = new RaftConsensusServiceImpl(raftNode);
server.registerService(raftConsensusService);
// 注册给Client调用的Raft服务
RaftClientService raftClientService = new RaftClientServiceImpl(raftNode);
server.registerService(raftClientService);
// 注册应用自己提供的服务
ExampleService exampleService = new ExampleServiceImpl(raftNode, stateMachine);
server.registerService(exampleService);
// 启动RPCServer,初始化Raft节点
server.start();
raftNode.init();

About

Raft Java implementation which is simple and easy to understand.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Java 95.6%
  • Shell 4.4%