第十五章 Seata处理分布式事务

Seata处理分布式事务

gitee:https://gitee.com/mougenan/springcloud_study.git

1. 分布式事务问题

例如: 用户购买商品的业务逻辑。整个业务逻辑由3个微服务提供克持: 仓储服务:对始定的向品扣除色储数量, 订单服务:根狐来购需求创建订单. 帐户服务:从用户帐户中扣除全额。

单体应用被拆分成微服务应用,原来的三个模块被拆分成三个独立的应用,分别使用三个独立的数据源,业务操作需要调用三个服务来完成。此时每个服务内部的数据—致性由本地事务来保证,但是全局的数据一致性问题没法保证。

一次业务操作需要跨多个数据源或需要跨多个系统进行远程调用,就会产生分布式事务问题

2. Seata简介

Seata是一款开源的分布式事务解决方案,致力于在微服务架构下提供高性能和简单易用的分布式事务服务。 官网:Seata | Seata

一个典型的分布式事务过程: 分布式事务处理过程的ID+三组件模型: 全局唯一的事务ID:Transaction ID XID 三组件概念: Transaction Coordinator(TC):事务协调器,维护全局事务的运行状态,负责协调并驱动全局事务的提交或回滚。 Transaction Manager(TM):控制全局事务的边界,负责开启一个全局事务,并最终发起全局提交或全局回滚的决议。 Resource Manager(RM):控制分支事务,负责分支注册、状态汇报,并接收事务协调器的指令,驱动分支(本地)事务的提交和回滚。 处理过程:

  1. TM向TC申请开启一个全局事务,全局事务创建成功并生成一个全局唯一的XID;

  2. XID在微服务调用链路的上下文中传播;

  3. RM向TC注册分支事务,将其纳入XID对应全局事务的管辖;

  4. TM向TC发起针对 XID的全局提交或回滚决议;

  5. TC调度XID下管辖的全部分支事务完成提交或回滚请求。

下载地址:Releases · apache/incubator-seata · GitHub 怎么使用:本地@Transactional、全局@GlobalTreansactional

3. Seata-Server

  1. 官网:Seata | Seata

  2. 下载版本0.9.0

  3. seata-server-0.9.0.zip解压到指定目录并修稿conf目录下的file.conf配置文件 修改:自定义事务组名称+事务日志存储模式为db+数据库连接信息

service {
  #vgroup->rgroup
  vgroup_mapping.my_test_tx_group = "fsp_txt_group"
  #only support single node
  default.grouplist = "127.0.0.1:8091"
  #degrade current not support
  enableDegrade = false
  #disable
  disable = false
  #unit ms,s,m,h,d represents milliseconds, seconds, minutes, hours, days, default permanent
  max.commit.retry.timeout = "-1"
  max.rollback.retry.timeout = "-1"
}
​
mode = "db"
​
  ## file store
  file {
    dir = "sessionStore"
​
    # branch session size , if exceeded first try compress lockkey, still exceeded throws exceptions
    max-branch-session-size = 16384
    # globe session size , if exceeded throws exceptions
    max-global-session-size = 512
    # file buffer size , if exceeded allocate new buffer
    file-write-buffer-cache-size = 16384
    # when recover batch read size
    session.reload.read_size = 100
    # async, sync
    flush-disk-mode = async
  }
​
  ## database store
  db {
    ## the implement of javax.sql.DataSource, such as DruidDataSource(druid)/BasicDataSource(dbcp) etc.
    datasource = "dbcp"
    ## mysql/oracle/h2/oceanbase etc.
    db-type = "mysql"
    driver-class-name = "com.mysql.jdbc.Driver"
    url = "jdbc:mysql://127.0.0.1:3306/seata"
    user = "root"
    password = "123456"
    min-conn = 1
    max-conn = 3
    global.table = "global_table"
    branch.table = "branch_table"
    lock-table = "lock_table"
    query-limit = 100
  }
  1. mysql5.7数据库建库seata,并在seata库中建表

  1. 修改seata-server-1.0.0\seata\conf目录下的registry.conf配置文件

  type = "nacos"
​
  nacos {
    serverAddr = "localhost:8848"
    namespace = ""
    cluster = "default"
  }
  1. 先启动Nacos端口8848

  2. 在启动seata-server

4. 订单/库存/账户业务数据库准备

业务说明:

这里我们会创建三个服务,一个订单服务,一个库存服务,一个账户服务。
​
当用户下单时,会在订单服务中创建一个订单,然后通过远程调用库存服务来扣减下单商品的库存,再通过远程调用账户服务来扣减用户账户里面的余额,
最后在订单服务中修改订单状态为已完成。
​
该操作跨越三个数据库,有两次远程调用,很明显会有分布式事务问题。

创建数据库:

seata_order:存储订单的数据库;
seata_storage:存储库存的数据库;
seata_acdount:存储账户信息的数据库。

创建业务表:

seata_order库下建t_order表
    CREATE TABLE t_order (
    `id BIGINT(11)NOT NULL AUTO_INCREMENT PRIMARY KEY,
    'user_id` BIGINT(11) DEFAULT NULL COMMENT "用户id",
    `product_id` BIGINT(11)DEFAULT NULL COMMENT "产品id",
    `count` INT(11) DEFAULT NULL COMMENT "数量",
    `money` DECIMAL(11,0)DEFAULT NULL COMMENT "金额",
    `status` INT(1) DEFAULT NULL COMMENT "订单状态:0:创建手; 1:已完结"
    )ENGINE=INNODB AUTO_INCREMENT=7 DEFAULT CHARSET=utf8;
​
seata storage库下建t_storage表
    CREATE TABLE t_storage (
    `id` BIGINT(11)NOT NULL AUTO_INCREMENT PRIMARY KEY,
    `product_id` BIGINT(11) DEFAULT NULL COMMENT "产品id",
    `total` INT(11) DEFAULT NULL COMMENT "总库存",
    `used` INT(11) DEFAULT NULL COMMENT "已用库存",
    `residue` INT(11) DEFAULT NULL COMMENT "剩余库存"
    )ENGINE=INNODB AUTO_INCREMENT=2 DEFAULT CHARSET=utf8;
    
    INSERT INTO seata_storage.t_storage(`id`,`product_id`,`total`,`used`,`residue`)VALUES (1,'1',"100" , "0", "100");
    
seata_account库下建t_account表
    CREATE TABLE t_account (
`id`BIGINT(11)NOT NULL AUTO_INCREMENT PRIMARY KEY COMMENT "id",
`user_id` BIGINT(11) DEFAULT NULL COMMENT "用户id",
`total` DECIMAL(10,0) DEFAULT NULL COMMENT "总额度",
`usedT` DECIMAL(10,0) DEFAULT NULL COMMENT "已用余额",
`residue` DECIMAL(10,0)DEFAULT '0' COMMENT "剩余可用额度"
)ENGINE=INNODB AUTO_INCREMENT=2 DEFAULT CHARSET=utf8;
​
INSERT INTO seata_account.t_account(`id`,`user_id`,`total`,`usedT`,`residue`)
 VALUES (1,1,"1000","0","1000");

3个库分别创建回滚日志表:就是分别粘贴:\seata-server-0.9.0\seata\conf目录下的db_undo_log.sql。

最终效果:

5. 订单/库存/账户业务微服务准备

5.1 业务需求

下订单->减库存->扣余额->改(订单)状态

5.2 新建订单Order-Module
  1. 创建seata-order-service2001项目

  2. pom文件:

<dependency>
    <groupId>com.alibaba.cloud</groupId>
    <artifactId>spring-cloud-starter-alibaba-seata</artifactId>
    <exclusions>
        <exclusion>
            <groupId>seata-all</groupId>
            <artifactId>io.seata</artifactId>
        </exclusion>
    </exclusions>
</dependency>
<dependency>
    <groupId>io.seata</groupId>
    <artifactId>seata-all</artifactId>
    <version>0.9.0</version>
</dependency>
  1. application.yml

server:
  port: 2001
​
spring:
  application:
    name: seata-order-service
  cloud:
    alibaba:
      seata:
        tx-service-group: fsp_tx_group
    nacos:
      discovery:
        server-addr: localhost:8848
  datasource:
    driver-class-name: com.mysql.jdbc.Driver
    url: jdbc:mysql://localhost:3306/seata_order
    password: 123456
    username: root
feign:
  sentinel:
    enabled: true
logging:
  level:
    io:
      seata: info
mybatis:
  mapper-locations: classpath:mapper/*.xml
  1. file.conf

transport {
  # tcp udt unix-domain-socket
  type = "TCP"
  #NIO NATIVE
  server = "NIO"
  #enable heartbeat
  heartbeat = true
  #thread factory for netty
  thread-factory {
    boss-thread-prefix = "NettyBoss"
    worker-thread-prefix = "NettyServerNIOWorker"
    server-executor-thread-prefix = "NettyServerBizHandler"
    share-boss-worker = false
    client-selector-thread-prefix = "NettyClientSelector"
    client-selector-thread-size = 1
    client-worker-thread-prefix = "NettyClientWorkerThread"
    # netty boss thread size,will not be used for UDT
    boss-thread-size = 1
    #auto default pin or 8
    worker-thread-size = 8
  }
  shutdown {
    # when destroy server, wait seconds
    wait = 3
  }
  serialization = "seata"
  compressor = "none"
}
service {
  #vgroup->rgroup
  vgroup_mapping.fsp_txt_group = "default"
  #only support single node
  default.grouplist = "127.0.0.1:8091"
  #degrade current not support
  enableDegrade = false
  #disable
  disable = false
  #unit ms,s,m,h,d represents milliseconds, seconds, minutes, hours, days, default permanent
  max.commit.retry.timeout = "-1"
  max.rollback.retry.timeout = "-1"
}
​
client {
  async.commit.buffer.limit = 10000
  lock {
    retry.internal = 10
    retry.times = 30
  }
  report.retry.count = 5
  tm.commit.retry.count = 1
  tm.rollback.retry.count = 1
}
​
## transaction log store
store {
  ## store mode: file、db
  #mode = "file"
  mode = "db"
​
  ## file store
  file {
    dir = "sessionStore"
​
    # branch session size , if exceeded first try compress lockkey, still exceeded throws exceptions
    max-branch-session-size = 16384
    # globe session size , if exceeded throws exceptions
    max-global-session-size = 512
    # file buffer size , if exceeded allocate new buffer
    file-write-buffer-cache-size = 16384
    # when recover batch read size
    session.reload.read_size = 100
    # async, sync
    flush-disk-mode = async
  }
​
  ## database store
  db {
    ## the implement of javax.sql.DataSource, such as DruidDataSource(druid)/BasicDataSource(dbcp) etc.
    datasource = "dbcp"
    ## mysql/oracle/h2/oceanbase etc.
    db-type = "mysql"
    driver-class-name = "com.mysql.jdbc.Driver"
    url = "jdbc:mysql://127.0.0.1:3306/seata"
    user = "root"
    password = "123456"
    min-conn = 1
    max-conn = 3
    global.table = "global_table"
    branch.table = "branch_table"
    lock-table = "lock_table"
    query-limit = 100
  }
}
lock {
  ## the lock store mode: local、remote
  mode = "remote"
​
  local {
    ## store locks in user's database
  }
​
  remote {
    ## store locks in the seata's server
  }
}
recovery {
  #schedule committing retry period in milliseconds
  committing-retry-period = 1000
  #schedule asyn committing retry period in milliseconds
  asyn-committing-retry-period = 1000
  #schedule rollbacking retry period in milliseconds
  rollbacking-retry-period = 1000
  #schedule timeout retry period in milliseconds
  timeout-retry-period = 1000
}
​
transaction {
  undo.data.validation = true
  undo.log.serialization = "jackson"
  undo.log.save.days = 7
  #schedule delete expired undo_log in milliseconds
  undo.log.delete.period = 86400000
  undo.log.table = "undo_log"
}
​
## metrics settings
metrics {
  enabled = false
  registry-type = "compact"
  # multi exporters use comma divided
  exporter-list = "prometheus"
  exporter-prometheus-port = 9898
}
​
support {
  ## spring
  spring {
    # auto proxy the DataSource bean
    datasource.autoproxy = false
  }
}
  1. registry.conf

registry {
  # file 、nacos 、eureka、redis、zk、consul、etcd3、sofa
  #type = "file"
  type = "nacos"
​
  nacos {
    serverAddr = "localhost:8848"
    namespace = ""
    cluster = "default"
  }
  eureka {
    serviceUrl = "http://localhost:8761/eureka"
    application = "default"
    weight = "1"
  }
  redis {
    serverAddr = "localhost:6379"
    db = "0"
  }
  zk {
    cluster = "default"
    serverAddr = "127.0.0.1:2181"
    session.timeout = 6000
    connect.timeout = 2000
  }
  consul {
    cluster = "default"
    serverAddr = "127.0.0.1:8500"
  }
  etcd3 {
    cluster = "default"
    serverAddr = "http://localhost:2379"
  }
  sofa {
    serverAddr = "127.0.0.1:9603"
    application = "default"
    region = "DEFAULT_ZONE"
    datacenter = "DefaultDataCenter"
    cluster = "default"
    group = "SEATA_GROUP"
    addressWaitTime = "3000"
  }
  file {
    name = "file.conf"
  }
}
​
config {
  # file、nacos 、apollo、zk、consul、etcd3
  type = "file"
​
  nacos {
    serverAddr = "localhost"
    namespace = ""
  }
  consul {
    serverAddr = "127.0.0.1:8500"
  }
  apollo {
    app.id = "seata-server"
    apollo.meta = "http://192.168.1.204:8801"
  }
  zk {
    serverAddr = "127.0.0.1:2181"
    session.timeout = 6000
    connect.timeout = 2000
  }
  etcd3 {
    serverAddr = "http://localhost:2379"
  }
  file {
    name = "file.conf"
  }
}
​
  1. domain

@Data
@NoArgsConstructor
@AllArgsConstructor
public class Order {
    private Long id;
    private Long userId;
    private Long productId;
    private BigDecimal money;
    private Integer status;
}
@Data
@NoArgsConstructor
@AllArgsConstructor
public class CommonResult<T> {
    private Integer code;
    private String message;
    private T data;
    public CommonResult(Integer code,String message){
        this(code,message,null);
    }
}
  1. dao

@Mapper
public interface OrderDao {
    void create(Order order);
    void update(@Param("userId") Long userId,@Param("status")Integer status);
}
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
        "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.yicai.springcloud.dao.OrderDao">
​
    <resultMap id="BaseResultMap" type="com.yicai.springcloud.damain.Order">
        <id column="id" property="id" jdbcType="BIGINT"/>
        <result column="user_id" property="userId" jdbcType="BIGINT"/>
        <result column="product_id" property="productId" jdbcType="BIGINT"/>
        <result column="count" property="count" jdbcType="INTEGER"/>
        <result column="money" property="money" jdbcType="DECIMAL"/>
        <result column="status" property="status" jdbcType="INTEGER"/>
    </resultMap>
    <insert id="create">
        insert into t_order (id,user_id,product_id,`count`,money,status)
        values (null,#{userId},#{productId},#{count},#{money},0);
    </insert>
    <update id="update">
        update t_order set status=1 where user_id=#{userId} and status=#{status};
    </update>
</mapper>
  1. service

@Service
@Slf4j
public class OrderServiceImpl implements OrderService {
​
    @Resource
    private OrderDao orderDao;
    @Resource
    private StorageService storageService;
    @Resource
    private AccountService accountService;
​
    @Override
    public void create(Order order) {
        log.info("---------->开始新建订单");
        orderDao.create(order);
        log.info("---------->订单微服务开始调用库存,做扣减count");
        storageService.decrease(order.getProductId(),order.getCount());
        log.info("---------->订单微服务开始调用库存,做扣减end");
        log.info("---------->订单微服务开始调用账户,做扣减money");
        accountService.decrease(order.getUserId(),order.getMoney());
        log.info("---------->订单微服务开始调用账户,做扣减end");
        log.info("---------->修改订单状态开始");
        orderDao.update(order.getUserId(),0);
        log.info("---------->修改订单状态结束");
        log.info("---------->下订单结束了!");
    }
}
  1. controller

@RestController
@Slf4j
public class OrderController {
​
    @Resource
    public OrderService orderService;
​
    @GetMapping("/order/create")
    public CommonResult create(Order order){
        orderService.create(order);
        return new CommonResult(200,"订单创建成功");
    }
}
  1. config

@Configuration
public class DataSourceProxyConfig {
​
    @Value("${mybatis.mapper-locations")
    private String mapperLocations;
​
    @Bean
    @ConfigurationProperties(prefix = "spring.datasource")
    public DataSource druidDataSource(){
        return new DruidDataSource();
    }
​
    @Bean
    public DataSourceProxy dataSourceProxy(DataSource dataSource){
        return new DataSourceProxy(dataSource);
    }
​
    @Bean
    public SqlSessionFactory sqlSessionFactoryBean(DataSourceProxy dataSourceProxy) throws Exception {
        SqlSessionFactoryBean sqlSessionFactoryBean = new SqlSessionFactoryBean();
        sqlSessionFactoryBean.setDataSource(dataSourceProxy);
        sqlSessionFactoryBean.setMapperLocations(new PathMatchingResourcePatternResolver().getResources(mapperLocations));
        sqlSessionFactoryBean.setTransactionFactory(new SpringManagedTransactionFactory());
        return sqlSessionFactoryBean.getObject();
    }
}
  1. 主启动

@SpringBootApplication(exclude = DataSourceAutoConfiguration.class)
@EnableDiscoveryClient
@EnableFeignClients
public class OrderMain2001 {
    public static void main(String[] args) {
        SpringApplication.run(OrderMain2001.class,args);
    }
}
5.3 新建库存Storage -Module
  1. 创建seata-storage-service2002项目

  2. pom.xml文件与2001一样

  3. application.yml基本一样

  4. file.conf和register.conf一样

  5. domain

@AllArgsConstructor
@NoArgsConstructor
@Data
public class Storage {
    private Long id;
    private Long productId;
    private Integer total;
    private Integer used;
    private Integer residue;
}
  1. dao

@Mapper
public interface StorageDao {
    void decrease(@Param("productId")Long productId,@Param("count")Integer count);
}
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
        "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.yicai.springcloud.dao.StorageDao">
​
    <resultMap id="BaseResultMap" type="com.yicai.springcloud.domain.Storage">
        <id column="id" property="id" jdbcType="BIGINT"/>
        <result column="product_id" property="productId" jdbcType="INTEGER"/>
        <result column="total" property="total" jdbcType="INTEGER"/>
        <result column="used" property="used" jdbcType="INTEGER"/>
        <result column="residue" property="residue" jdbcType="INTEGER"/>
    </resultMap>
​
    <update id="decrease">
        update t_storage
        set used=used+#{count},residue=residue-#{count}
        where product_id=#{productId};
    </update>
</mapper>
  1. service

public interface StorageService {
    void decrease(Long productId, Integer count);
}
@Service
public class StorageServiceImpl implements StorageService {
​
    private static final Logger LOGGER= LoggerFactory.getLogger(StorageServiceImpl.class);
​
    @Resource
    private StorageDao storageDao;
​
    @Override
    public void decrease(Long productId, Integer count) {
        LOGGER.info("--------->storage-service中扣减库存开始");
        storageDao.decrease(productId,count);
        LOGGER.info("------------->storage-service中扣减库存结束");
    }
}
  1. config与2001一样

  2. controller

@RestController
@Slf4j
public class StorageController {
    
    @Autowired
    private StorageService storageService;
    
    @RequestMapping("/storage/decrease")
    public CommonResult decrease(Long productId,Integer count){
        storageService.decrease(productId,count);
        return new CommonResult(200,"扣减库存成功!");
    }
}
5.4 新建账户Account-Module
  1. 创建seata-account-service2003项目

  2. pom.xml文件与2001一样

  3. application.yml基本一样

  4. file.conf和register.conf一样

  5. domain

@Data
@AllArgsConstructor
@NoArgsConstructor
public class Account {
    private Long id;
    private Long userId;
    private BigDecimal total;
    private BigDecimal usedT;
    private BigDecimal residue;
}
  1. dao

@Mapper
public interface AccountDao {
    void decrease(@Param("userId")Long userId, @Param("money")BigDecimal money);
}
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mapper PUBLIC "-//mybatis.org//DTD Mapper 3.0//EN"
        "http://mybatis.org/dtd/mybatis-3-mapper.dtd">
<mapper namespace="com.yicai.springcloud.dao.AccountDao">
​
    <resultMap id="BaseResultMap" type="com.yicai.springcloud.domain.Account">
        <id column="id" property="id" jdbcType="BIGINT"/>
        <result column="user_id" property="userId" jdbcType="BIGINT"/>
        <result column="total" property="total" jdbcType="DECIMAL"/>
        <result column="usedT" property="usedT" jdbcType="DECIMAL"/>
        <result column="residue" property="residue" jdbcType="DECIMAL"/>
    </resultMap>
​
    <update id="decrease">
        update t_account
        set residue = residue-#{money},usedT = usedT + #{money} 
        where user_id=#{userId};
    </update>
</mapper>
  1. service

public interface AccountService {
    void decrease(Long userId, BigDecimal money);
}
@Service
public class AccountServiceImpl implements AccountService {
​
    private static final Logger LOGGER= LoggerFactory.getLogger(AccountServiceImpl.class);
    
    @Resource
    private AccountDao accountDao;
​
    @Override
    public void decrease(Long userId, BigDecimal money) {
        LOGGER.info("--------->account-service中扣减库存开始");
        accountDao.decrease(userId,money);
        LOGGER.info("------------->account-service中扣减库存结束");
    }
}
  1. config与2001一样

  2. controller

@RestController
@Slf4j
public class AccountController {
    
    @Resource
    AccountService accountService;
    
    @RequestMapping("/account/decrease")
    public CommonResult decrease(@RequestParam("userId")Long userId, @RequestParam("money")BigDecimal money){
        accountService.decrease(userId,money);
        return new CommonResult(200,"扣减账余额成功!");
    }
}
5.5 测试

正常下单:

超时异常:

@Service
public class AccountServiceImpl implements AccountService {
​
    private static final Logger LOGGER= LoggerFactory.getLogger(AccountServiceImpl.class);
​
    @Resource
    private AccountDao accountDao;
​
    @Override
    public void decrease(Long userId, BigDecimal money) {
        LOGGER.info("--------->account-service中扣减库存开始");
        try{
            TimeUnit.SECONDS.sleep(20000);
        }catch (InterruptedException e){
            e.printStackTrace();
        }
        accountDao.decrease(userId,money);
        LOGGER.info("------------->account-service中扣减库存结束");
    }
}

当库存和账户金额扣减后,订单状态并没有设置为已经完成,没有从零改为1而且由于feign的重试机制,账户余额还有可能被多次扣减。

@GlobalTransactional测试

@Override
    @GlobalTransactional(name = "fsp-create-order",rollbackFor = Exception.class)
    public void create(Order order) {
        log.info("---------->开始新建订单");
        orderDao.create(order);
        log.info("---------->订单微服务开始调用库存,做扣减count");
        storageService.decrease(order.getProductId(),order.getCount());
        log.info("---------->订单微服务开始调用库存,做扣减end");
        log.info("---------->订单微服务开始调用账户,做扣减money");
        accountService.decrease(order.getUserId(),order.getMoney());
        log.info("---------->订单微服务开始调用账户,做扣减end");
        log.info("---------->修改订单状态开始");
        orderDao.update(order.getUserId(),0);
        log.info("---------->修改订单状态结束");
        log.info("---------->下订单结束了!");
    }

下单后数据库教据并没有任何改变;记录都添加不进来少。

5.6 Seata之原理简介

Seata:2019年1月份蚂蚁金服和阿里巴巴共同开源的分布式事务解决方案 Simple Extensible Autonomous Transaction Architecture,简单可扩展自治事务框架。

TM开启分布式事务(TM向TC注册全局事务记录); 按业务场景,编排数据库、服务等事务内资源(RM向TC汇报资源准备状态); TM结束分布式事务,事务一阶段结束(TM通知TC提交/回滚分布式事务); TC汇总事务信息,决定分布式事务是提交还是回滚; TC通知所有RM提交/回滚资源,事务二阶段结束。

在一阶段,Seata 会拦截“业务SQL”, 1.解析SQL语义,找到“业务SQL”要更新的业务数据,在业务数据被更新前,将其保存成“before image”, 2.执行“业务SQL”更新业务数据,在业务数据更新之后, 3.其保存成“after image”,最后生成行锁。 以上操作全部在一个数据库事务内完成,这样保证了一阶段操作的原子性。

二阶段如是顺利提交的话,因为“业务SQL”在一阶段已经提交至数据库,所以Seata框架只需将一阶段保存的快照数据和行锁删掉,完成数据清理即可。

二阶段回滚:二阶段如果是回滚的话,Seata就需要回滚一阶段已经执行的“业务SQL”,还原业务数据。回滚方式便是用“before image”还原业务数据;但在还原前要首先要校验脏写,对比“数据库当前业务数据”和“after image"如果两份数据完全一致就说明没有脏写,可以还原业务数据,如果不一致就说明有脏写,出现脏写就需要转人工处理。

总结:Seata的 AT、XA模式都是基于全局事务实现的,在高并发的场景下会出现获取全局锁异常,因此这两种模式都不适用高并发场景;Seata TCC模式性能比AT模式的好一点,但是并发量大于100的话还是不适合;如果基本没有什么并发量的话,可以选择AT模式;并发量在一百内的话可以使用TCC模式。

  • 20
    点赞
  • 19
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 0
    评论

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

我愿为一粒沙

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值