分布式事务Seata使用及其原理剖析 - Seata快速实战

携手创作，共同成长！这是我参与「掘金日新计划 · 8 月更文挑战」的第12天，点击查看活动详情

Seata快速开始

1 Seata Server（TC）环境搭建

https://seata.io/zh-cn/docs/ops/deploy-guide-beginner.html

Server端存储模式（store.mode）支持三种：

• file：单机模式，全局事务会话信息内存中读写并持久化本地文件root.data，性能较高
• db：高可用模式，全局事务会话信息通过db共享，相应性能差些
• redis：Seata-Server 1.3及以上版本支持,性能较高,存在事务信息丢失风险,请提前配置适合当前场景的redis持久化配置

资源目录：github.com/seata/seata…

• client

存放client端sql脚本，参数配置

• config-center

各个配置中心参数导入脚本，config.txt(包含server和client，原名nacos-config.txt)为通用参数文件

• server

server端数据库脚本及各个容器配置

db存储模式+Nacos(注册&配置中心)部署

步骤一：下载安装包

github.com/seata/seata…

步骤二：建表(仅db模式)

全局事务会话信息由3块内容构成，全局事务-->分支事务-->全局锁，对应表global_table、branch_table、lock_table

创建数据库seata，执行sql脚本，文件在script/server/db/mysql.sql（seata源码）中

步骤三：修改store.mode

启动包: seata-->conf-->file.conf，修改store.mode="db"

源码: 根目录-->seata-server-->resources-->file.conf，修改store.mode="db"

步骤四：修改数据库连接

启动包: seata-->conf-->file.conf，修改store.db相关属性。

源码: 根目录-->seata-server-->resources-->file.conf，修改store.db相关属性。

此时可以调到步骤七：直接启动Seata Server，注册中心和配置中心都是file

步骤五：配置Nacos注册中心

将Seata Server注册到Nacos，修改conf目录下的registry.conf配置

然后启动注册中心Nacos Server

#进入Nacos安装目录，linux单机启动
bin/startup.sh -m standalone
# windows单机启动
bin/startup.bat

步骤六：配置Nacos配置中心

注意：如果配置了seata server使用nacos作为配置中心，则配置信息会从nacos读取，file.conf可以不用配置。 客户端配置registry.conf使用nacos时也要注意group要和seata server中的group一致，默认group是"DEFAULT_GROUP"

获取/seata/script/config-center/config.txt，修改配置信息

配置事务分组， 要与客户端配置的事务分组一致

（客户端properties配置：spring.cloud.alibaba.seata.tx‐service‐group=my_test_tx_group）

配置参数同步到Nacos

shell:

sh ${SEATAPATH}/script/config-center/nacos/nacos-config.sh -h localhost -p 8848 -g SEATA_GROUP -t 5a3c7d6c-f497-4d68-a71a-2e5e3340b3ca

参数说明：

-h: host，默认值 localhost

-p: port，默认值 8848

-g: 配置分组，默认值为 'SEATA_GROUP'

-t: 租户信息，对应 Nacos 的命名空间ID字段, 默认值为空 ''

精简配置

service.vgroupMapping.my_test_tx_group=default
service.default.grouplist=127.0.0.1:8091
service.enableDegrade=false
service.disableGlobalTransaction=false
store.mode=db
store.db.datasource=druid
store.db.dbType=mysql
store.db.driverClassName=com.mysql.jdbc.Driver
store.db.url=jdbc:mysql://127.0.0.1:3306/seata?useUnicode=true
store.db.user=root
store.db.password=root
store.db.minConn=5
store.db.maxConn=30
store.db.globalTable=global_table
store.db.branchTable=branch_table
store.db.queryLimit=100
store.db.lockTable=lock_table
store.db.maxWait=5000

步骤七：启动Seata Server

• 源码启动: 执行server模块下io.seata.server.Server.java的main方法
• 命令启动: bin/seata-server.sh -h 127.0.0.1 -p 8091 -m db -n 1 -e test

启动Seata Server

bin/seata-server.sh
bin/seata-server.sh -p 80

启动成功，默认端口8091

在注册中心中可以查看到seata-server注册成功

2 Seata Client快速开始

编程式事务实现（GlobalTransaction API）

Demo：seata-samples/api

客户端环境配置

1. 修改jdbc.properties配置

2. registry.conf中指定registry.type="file" , config.type="file"

基于GlobalTransaction API的实现

public static void main(String[] args) throws SQLException, TransactionException, InterruptedException {

    String userId = "U100001";
    String commodityCode = "C00321";
    int commodityCount = 100;
    int money = 999;
    AccountService accountService = new AccountServiceImpl();
    StorageService storageService = new StorageServiceImpl();
    OrderService orderService = new OrderServiceImpl();
    orderService.setAccountService(accountService);

    //reset data  重置数据
    accountService.reset(userId, String.valueOf(money));
    storageService.reset(commodityCode, String.valueOf(commodityCount));
    orderService.reset(null, null);

    //init seata; only once
    String applicationId = "api";
    String txServiceGroup = "my_test_tx_group";
    TMClient.init(applicationId, txServiceGroup);
    RMClient.init(applicationId, txServiceGroup);

    //trx  开启全局事务
    GlobalTransaction tx = GlobalTransactionContext.getCurrentOrCreate();
    try {
        tx.begin(60000, "testBiz");
        System.out.println("begin trx, xid is " + tx.getXid());

        //biz operate 3 dataSources
        //set >=5 will be rollback(200*5>999) else will be commit
        int opCount = 5;
        // 扣减库存
        storageService.deduct(commodityCode, opCount);
        // 创建订单 ，扣款 money = opCount * 200
        orderService.create(userId, commodityCode, opCount);

        //check data if negative
        boolean needCommit = ((StorageServiceImpl)storageService).validNegativeCheck("count", commodityCode)
            && ((AccountServiceImpl)accountService).validNegativeCheck("money", userId);

        //if data negative rollback else commit
        if (needCommit) {
            tx.commit();
        } else {
            System.out.println("rollback trx, cause: data negative, xid is " + tx.getXid());
            tx.rollback();
        }
    } catch (Exception exx) {
        System.out.println("rollback trx, cause: " + exx.getMessage() + " , xid is " + tx.getXid());
        tx.rollback();
        throw exx;
    }
    TimeUnit.SECONDS.sleep(10);

}

声明式事务实现（@GlobalTransactional）

业务场景：

用户下单，整个业务逻辑由三个服务构成：

• 仓储服务：对给定的商品扣除库存数量。
• 订单服务：根据采购需求创建订单。
• 帐户服务：从用户帐户中扣除余额。

多数据源场景

1. 启动seata server服务，指定registry.type="file" , config.type="file"

2. 客户端应用接入seata配置

   1）配置多数据源

​ 客户端支持多数据源，yml中添加多数据源jdbc配置

# Order
spring.datasource.order.url=jdbc:mysql://localhost:3306/seata_order?useUnicode=true&characterEncoding=utf8&allowMultiQueries=true&useSSL=false&serverTimezone=UTC
spring.datasource.order.username=root
spring.datasource.order.password=root
spring.datasource.order.driver-class-name=com.mysql.cj.jdbc.Driver
# Storage
spring.datasource.storage.url=jdbc:mysql://localhost:3306/seata_storage?useUnicode=true&characterEncoding=utf8&allowMultiQueries=true&useSSL=false&serverTimezone=UTC
spring.datasource.storage.username=root
spring.datasource.storage.password=root
spring.datasource.storage.driver-class-name=com.mysql.cj.jdbc.Driver
# Account
spring.datasource.account.url=jdbc:mysql://localhost:3306/seata_account?useUnicode=true&characterEncoding=utf8&allowMultiQueries=true&useSSL=false&serverTimezone=UTC
spring.datasource.account.username=root
spring.datasource.account.password=root
spring.datasource.account.driver-class-name=com.mysq

​

2）配置多数据源代理，并支持动态切换数据源

@Configuration
@MapperScan("com.tuling.mutiple.datasource.mapper")
public class DataSourceProxyConfig {

    @Bean("originOrder")
    @ConfigurationProperties(prefix = "spring.datasource.order")
    public DataSource dataSourceMaster() {
        return new DruidDataSource();
    }

    @Bean("originStorage")
    @ConfigurationProperties(prefix = "spring.datasource.storage")
    public DataSource dataSourceStorage() {
        return new DruidDataSource();
    }

    @Bean("originAccount")
    @ConfigurationProperties(prefix = "spring.datasource.account")
    public DataSource dataSourceAccount() {
        return new DruidDataSource();
    }

    @Bean(name = "order")
    public DataSourceProxy masterDataSourceProxy(@Qualifier("originOrder") DataSource dataSource) {
        return new DataSourceProxy(dataSource);
    }

    @Bean(name = "storage")
    public DataSourceProxy storageDataSourceProxy(@Qualifier("originStorage") DataSource dataSource) {
        return new DataSourceProxy(dataSource);
    }

    @Bean(name = "account")
    public DataSourceProxy payDataSourceProxy(@Qualifier("originAccount") DataSource dataSource) {
        return new DataSourceProxy(dataSource);
    }

    @Bean("dynamicDataSource")
    public DataSource dynamicDataSource(@Qualifier("order") DataSource dataSourceOrder,
                                        @Qualifier("storage") DataSource dataSourceStorage,
                                        @Qualifier("account") DataSource dataSourcePay) {

        DynamicRoutingDataSource dynamicRoutingDataSource = new DynamicRoutingDataSource();

        // 数据源的集合
        Map<Object, Object> dataSourceMap = new HashMap<>(3);
        dataSourceMap.put(DataSourceKey.ORDER.name(), dataSourceOrder);
        dataSourceMap.put(DataSourceKey.STORAGE.name(), dataSourceStorage);
        dataSourceMap.put(DataSourceKey.ACCOUNT.name(), dataSourcePay);

        dynamicRoutingDataSource.setDefaultTargetDataSource(dataSourceOrder);
        dynamicRoutingDataSource.setTargetDataSources(dataSourceMap);

        DynamicDataSourceContextHolder.getDataSourceKeys().addAll(dataSourceMap.keySet());

        return dynamicRoutingDataSource;
    }

    @Bean
    @ConfigurationProperties(prefix = "mybatis")
    public SqlSessionFactoryBean sqlSessionFactoryBean(@Qualifier("dynamicDataSource") DataSource dataSource) {
        SqlSessionFactoryBean sqlSessionFactoryBean = new SqlSessionFactoryBean();
        sqlSessionFactoryBean.setDataSource(dataSource);
        return sqlSessionFactoryBean;
    }

}

@Slf4j
public class DynamicRoutingDataSource extends AbstractRoutingDataSource {

    @Override
    protected Object determineCurrentLookupKey() {
        log.info("当前数据源 [{}]", DynamicDataSourceContextHolder.getDataSourceKey());
        return DynamicDataSourceContextHolder.getDataSourceKey();
    }
}

public class DynamicDataSourceContextHolder {

    private static final ThreadLocal<String> CONTEXT_HOLDER = ThreadLocal.withInitial(DataSourceKey.ORDER::name);

    private static List<Object> dataSourceKeys = new ArrayList<>();

    public static void setDataSourceKey(DataSourceKey key) {
        CONTEXT_HOLDER.set(key.name());
    }

    public static String getDataSourceKey() {
        return CONTEXT_HOLDER.get();
    }

    public static void clearDataSourceKey() {
        CONTEXT_HOLDER.remove();
    }

    public static List<Object> getDataSourceKeys() {
        return dataSourceKeys;
    }
}

3）接入seata配置

registry.conf中指定registry.type="file" , config.type="file" ,对应seata-server的registry.conf配置相同

registry {
  # file 、nacos 、eureka、redis、zk、consul、etcd3、sofa
  type = "file"

  file {
    name = "file.conf"
  }
}

config {
  # file、nacos 、apollo、zk、consul、etcd3、springCloudConfig
  type = "file"

  file {
    name = "file.conf"
  }
}

4）指定seata事务分组，用于获取seata server服务实例

# Seata事务分组  从file.conf获取service.vgroupMapping.my_test_tx_group的集群名称default，用于确定seata server的服务实例
spring.cloud.alibaba.seata.tx-service-group=my_test_tx_group

5）OrderServiceImpl作为发起者配置@GlobalTransactional注解

@Override
//@Transactional
@GlobalTransactional(name="createOrder")
public Order saveOrder(OrderVo orderVo){
    log.info("=============用户下单=================");
    //切换数据源
    DynamicDataSourceContextHolder.setDataSourceKey(DataSourceKey.ORDER);
    log.info("当前 XID: {}", RootContext.getXID());
    
    // 保存订单
    Order order = new Order();
    order.setUserId(orderVo.getUserId());
    order.setCommodityCode(orderVo.getCommodityCode());
    order.setCount(orderVo.getCount());
    order.setMoney(orderVo.getMoney());
    order.setStatus(OrderStatus.INIT.getValue());

    Integer saveOrderRecord = orderMapper.insert(order);
    log.info("保存订单{}", saveOrderRecord > 0 ? "成功" : "失败");
    
    //扣减库存
    storageService.deduct(orderVo.getCommodityCode(),orderVo.getCount());
    
    //扣减余额
    accountService.debit(orderVo.getUserId(),orderVo.getMoney());

    log.info("=============更新订单状态=================");
    //切换数据源
    DynamicDataSourceContextHolder.setDataSourceKey(DataSourceKey.ORDER);
    //更新订单
    Integer updateOrderRecord = orderMapper.updateOrderStatus(order.getId(),OrderStatus.SUCCESS.getValue());
    log.info("更新订单id:{} {}", order.getId(), updateOrderRecord > 0 ? "成功" : "失败");
    
    return order;
    
}

测试成功场景

调用 /order/createOrder 接口，将 money 设置为 10，此时余额为 20，可以下单成功

测试失败场景

设置 money 为 100，此时余额不足，会下单失败，account-service会抛出异常，事务会回滚

接入微服务应用

业务场景：

用户下单，整个业务逻辑由三个微服务构成：

• 仓储服务：对给定的商品扣除库存数量。
• 订单服务：根据采购需求创建订单。
• 帐户服务：从用户帐户中扣除余额。

1）启动Seata server端，Seata server使用nacos作为配置中心和注册中心

2）配置微服务整合seata

第一步：添加pom依赖

<!-- seata-->
<dependency>
    <groupId>com.alibaba.cloud</groupId>
    <artifactId>spring-cloud-starter-alibaba-seata</artifactId>
    <exclusions>
        <exclusion>
            <groupId>io.seata</groupId>
            <artifactId>seata-all</artifactId>
        </exclusion>
    </exclusions>
</dependency>
<dependency>
    <groupId>io.seata</groupId>
    <artifactId>seata-all</artifactId>
    <version>1.4.0</version>
</dependency>

第二步： 微服务对应数据库中添加undo_log表

CREATE TABLE `undo_log` (
  `id` bigint(20) NOT NULL AUTO_INCREMENT,
  `branch_id` bigint(20) NOT NULL,
  `xid` varchar(100) NOT NULL,
  `context` varchar(128) NOT NULL,
  `rollback_info` longblob NOT NULL,
  `log_status` int(11) NOT NULL,
  `log_created` datetime NOT NULL,
  `log_modified` datetime NOT NULL,
  PRIMARY KEY (`id`),
  UNIQUE KEY `ux_undo_log` (`xid`,`branch_id`)
) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8;

第三步：添加代理数据源配置，配置DataSourceProxy

/**
 * @author Fox
 *
 * 需要用到分布式事务的微服务都需要使用seata DataSourceProxy代理自己的数据源
 */
@Configuration
@MapperScan("com.tuling.datasource.mapper")
public class MybatisConfig {
    
    /**
     * 从配置文件获取属性构造datasource，注意前缀，这里用的是druid，根据自己情况配置,
     * 原生datasource前缀取"spring.datasource"
     *
     * @return
     */
    @Bean
    @ConfigurationProperties(prefix = "spring.datasource.druid")
    public DataSource druidDataSource() {
        DruidDataSource druidDataSource = new DruidDataSource();
        return druidDataSource;
    }
    
    /**
     * 构造datasource代理对象，替换原来的datasource
     * @param druidDataSource
     * @return
     */
    @Primary
    @Bean("dataSource")
    public DataSourceProxy dataSourceProxy(DataSource druidDataSource) {
        return new DataSourceProxy(druidDataSource);
    }
    
    
    @Bean(name = "sqlSessionFactory")
    public SqlSessionFactory sqlSessionFactoryBean(DataSourceProxy dataSourceProxy) throws Exception {
        SqlSessionFactoryBean factoryBean = new SqlSessionFactoryBean();
        //设置代理数据源
        factoryBean.setDataSource(dataSourceProxy);
        ResourcePatternResolver resolver = new PathMatchingResourcePatternResolver();
        factoryBean.setMapperLocations(resolver.getResources("classpath*:mybatis/**/*-mapper.xml"));
        
        org.apache.ibatis.session.Configuration configuration=new org.apache.ibatis.session.Configuration();
        //使用jdbc的getGeneratedKeys获取数据库自增主键值
        configuration.setUseGeneratedKeys(true);
        //使用列别名替换列名
        configuration.setUseColumnLabel(true);
        //自动使用驼峰命名属性映射字段，如userId ---> user_id
        configuration.setMapUnderscoreToCamelCase(true);
        factoryBean.setConfiguration(configuration);
        
        return factoryBean.getObject();
    }
    
}

第四步：启动类上剔除DataSourceAutoConfiguration，用于解决数据源的循环依赖问题

@SpringBootApplication(scanBasePackages = "com.tuling",exclude = DataSourceAutoConfiguration.class)
@EnableFeignClients
public class OrderServiceApplication {

    public static void main(String[] args) {
        SpringApplication.run(OrderServiceApplication.class, args);
    }

}

第五步：修改register.conf,配置nacos作为registry.type&config.type，对应seata server也使用nacos

注意：需要指定group = "SEATA_GROUP"，因为Seata Server端指定了group = "SEATA_GROUP" ，必须保证一致

registry {
  # file 、nacos 、eureka、redis、zk、consul、etcd3、sofa
  type = "nacos"
  
  nacos {
      serverAddr = "localhost"
      namespace = ""
      cluster = "default"
      group = "SEATA_GROUP"
  }
}
config {
  # file、nacos 、apollo、zk、consul、etcd3、springCloudConfig
  type = "nacos"

  nacos {
    serverAddr = "localhost"
    namespace = ""
    group = "SEATA_GROUP"
  }
}

如果出现这种问题：

一般大多数情况下都是因为配置不匹配导致的：

1. 检查现在使用的seata服务和项目maven中seata的版本是否一致

2. 检查tx-service-group，nacos.cluster，nacos.group参数是否和Seata Server中的配置一致

跟踪源码：seata/discover包下实现了RegistryService#lookup，用来获取服务列表

NacosRegistryServiceImpl#lookup
》String clusterName = getServiceGroup(key);  #获取seata server集群名称
》List<Instance> firstAllInstances = getNamingInstance().getAllInstances(getServiceName(), getServiceGroup(), clusters)

第六步：修改application.yml配置

配置seata 服务事务分组，要与服务端nacos配置中心中service.vgroup_mapping的后缀对应

server:
  port: 8020

spring:
  application:
    name: order-service
  cloud:
    nacos:
      discovery:
        server-addr: 127.0.0.1:8848
    alibaba:
      seata:
        tx-service-group:
          my_test_tx_group # seata 服务事务分组

  datasource:
    type: com.alibaba.druid.pool.DruidDataSource
    druid:
      driver-class-name: com.mysql.cj.jdbc.Driver
      url: jdbc:mysql://localhost:3306/seata_order?useUnicode=true&characterEncoding=UTF-8&serverTimezone=Asia/Shanghai
      username: root
      password: root
      initial-size: 10
      max-active: 100
      min-idle: 10
      max-wait: 60000
      pool-prepared-statements: true
      max-pool-prepared-statement-per-connection-size: 20
      time-between-eviction-runs-millis: 60000
      min-evictable-idle-time-millis: 300000
      test-while-idle: true
      test-on-borrow: false
      test-on-return: false
      stat-view-servlet:
        enabled: true
        url-pattern: /druid/*
      filter:
        stat:
          log-slow-sql: true
          slow-sql-millis: 1000
          merge-sql: false
        wall:
          config:
            multi-statement-allow: true

第七步：微服务发起者（TM 方）需要添加@GlobalTransactional注解

@Override
//@Transactional
@GlobalTransactional(name="createOrder")
public Order saveOrder(OrderVo orderVo){
    log.info("=============用户下单=================");
    log.info("当前 XID: {}", RootContext.getXID());
    
    // 保存订单
    Order order = new Order();
    order.setUserId(orderVo.getUserId());
    order.setCommodityCode(orderVo.getCommodityCode());
    order.setCount(orderVo.getCount());
    order.setMoney(orderVo.getMoney());
    order.setStatus(OrderStatus.INIT.getValue());

    Integer saveOrderRecord = orderMapper.insert(order);
    log.info("保存订单{}", saveOrderRecord > 0 ? "成功" : "失败");
    
    //扣减库存
    storageFeignService.deduct(orderVo.getCommodityCode(),orderVo.getCount());
    
    //扣减余额
    accountFeignService.debit(orderVo.getUserId(),orderVo.getMoney());

    //更新订单
    Integer updateOrderRecord = orderMapper.updateOrderStatus(order.getId(),OrderStatus.SUCCESS.getValue());
    log.info("更新订单id:{} {}", order.getId(), updateOrderRecord > 0 ? "成功" : "失败");
    
    return order;
    
}

测试

分布式事务成功，模拟正常下单、扣库存，扣余额

分布式事务失败，模拟下单扣库存成功、扣余额失败，事务是否回滚