- Dts是一款高性能、高可靠、接入简单的分布式事务解决方案,用于解决分布式环境下的事务一致性问题
- 支持Spring Cloud下分布式事务
- 避免脏读、幻读
- 基于2PC提交方式，接入简单,只要配置一个注解就可以开启分布式事务
- 类似于阿里Gts和蚂蚁金服DTX的实现
- 支持集群及HA
dts-common: 基础包，包含client、resource与server的协议
dts-core: client、resource的实现
dts-remoting: 网络实现，基于netty实现client、reousrce于server之间的交互
dts-server: 事务协调器，基于netty
dts-ops: dts管理控制台
引入二方库坐标
<dependency>
<groupId>com.bkjk.platform.summerframework</groupId>
<artifactId>springcloud-starter-dts</artifactId>
<version>2.0.0-SNAPSHOT</version>
</dependency>
在服务调用方、服务提供方执行以下SQL，来保存redo/undo的日志及行锁信息
DROP TABLE IF EXISTS `dts_branch_info`;
CREATE TABLE `dts_branch_info` (
`id` bigint(11) NOT NULL AUTO_INCREMENT,
`trans_id` bigint(20) NOT NULL COMMENT '事务号',
`branch_id` bigint(20) NOT NULL COMMENT '分支事务号',
`log_info` longblob COMMENT 'undo/redo log',
`gmt_create` datetime DEFAULT NULL COMMENT '创建时间',
`gmt_modified` datetime DEFAULT NULL COMMENT '修改时间',
`status` int(10) DEFAULT NULL COMMENT '事务状态',
`instance_id` varchar(250) DEFAULT NULL COMMENT '服务实例Id',
PRIMARY KEY (`id`),
KEY `branch_id` (`branch_id`,`trans_id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1 COMMENT='redo/undo备份日志表';
DROP TABLE IF EXISTS `dts_row_lock`;
CREATE TABLE `dts_row_lock` (
`id` bigint(11) NOT NULL AUTO_INCREMENT,
`branch_id` bigint(20) NOT NULL COMMENT '分支事务号',
`trans_id` bigint(20) NOT NULL COMMENT '主事务号',
`table_name` varchar(64) NOT NULL COMMENT '表名称',
`row_key` varchar(250) NOT NULL COMMENT '行唯一key',
`instance_id` varchar(250) NOT NULL COMMENT '实例id',
`gmt_create` datetime NOT NULL COMMENT '创建时间',
`gmt_modified` datetime NOT NULL COMMENT '修改时间',
PRIMARY KEY (`id`),
UNIQUE KEY `idx_row_lock_id` (`table_name`,`row_key`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COMMENT='行锁';
服务调用方：在服务调用不同的接口添加@DtsTransaction注解，将两个服务调用纳入整个分布式事务管理
@DtsTransaction
public String rollbackDelete(@RequestBody StudentDo studentDo) {
providerService.delete(studentDo);
providerService.runtimeException();
return "rollbackSelect";
}
服务提供方：在服务提供方使用Spring的@Transactional开启本地事务
@Transactional
public StudentDo update(@RequestBody StudentDo studentDo) {
studentMapper.updateById(studentDo);
return studentDo;
}
DTS Server部署
1:创建数据库，执行以下脚本
DROP TABLE IF EXISTS `dts_global_record`;
CREATE TABLE `dts_global_record` (
`trans_id` bigint(20) NOT NULL AUTO_INCREMENT,
`state` tinyint(1) NOT NULL COMMENT ' Begin(1),Committed(2),Rollbacked(3),CmmittedFailed(4),RollbackFailed(5),Commiting(6),Rollbacking(7);',
`gmt_created` datetime NOT NULL,
`gmt_modified` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
PRIMARY KEY (`trans_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_bin COMMENT='全局事务表';
DROP TABLE IF EXISTS `dts_branch_record`;
CREATE TABLE `dts_branch_record` (
`branch_id` bigint(20) NOT NULL AUTO_INCREMENT,
`trans_id` bigint(20) NOT NULL,
`resource_ip` varchar(200) COLLATE utf8_bin NOT NULL DEFAULT '',
`resource_info` varchar(200) COLLATE utf8_bin NOT NULL DEFAULT '',
`state` tinyint(1) NOT NULL COMMENT ' Begin(1), Success(2),Failed(3);',
`gmt_created` datetime NOT NULL,
`gmt_modified` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
PRIMARY KEY (`branch_id`),
KEY `tx_id` (`trans_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_bin COMMENT='事务分支记录表';
DROP TABLE IF EXISTS `dts_branch_error_log`;
CREATE TABLE `dts_branch_error_log` (
`branch_id` bigint(20) NOT NULL,
`trans_id` bigint(20) NOT NULL,
`resource_ip` varchar(200) COLLATE utf8_bin NOT NULL,
`resource_info` varchar(200) COLLATE utf8_bin NOT NULL,
`state` tinyint(1) NOT NULL DEFAULT '0',
`gmt_created` datetime NOT NULL,
`gmt_modified` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
`is_notify` tinyint(1) DEFAULT '0',
PRIMARY KEY (`branch_id`),
UNIQUE KEY `branch_id` (`branch_id`),
KEY `tx_id` (`trans_id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_bin COMMENT='事务分支错误日志表';
2：构建部署包
mvn clean install -Dmaven.test.skip=true