AT transaction model includes TM (Transaction Manager), RM (Resource Manager) and TC (Trans‐ action Coordinator). TC is an independent service that needs to be deployed separately. TM and RM are deployed together with user applications in the form of jar packages. They establish long connec‐ tions with TC and keep RPC throughout the transaction life cycle. The initiator of global transaction is TM, which mit/rollback is executed through TC coordination. A typical lifecycle of Seata managed distributed transaction: 1. TM asks TC to begin a new global transaction. TC generates a XID representing the global

TM (事务管理器)，RM (资源管理器) 和 TC (事务协调器)。TC 是一个独立部署的 服务，TM 和 RM 以 jar 包的方式同业务应用一同部署，它们同 TC 建立长连接，在整个事务生命周期内， 保持远程通信。TM 是全局事务的发起方，负责全局事务的开启，提交和回滚。RM 是全局事务的参与者， 负责分支事务的执行结果上报，并且通过 TC 的协调进行分支事务的提交和回滚。 Seata Seata 管理的分布式事务的典型生命周期： 1. TM 要求 TC 开始一个全新的全局事务。TC 生成一个代表该全局事务的 XID。 2. XID 贯穿于微服务的整个调用链。 3. 作为该 XID 对应到的 TC 下的全局事务的一部分，RM 注册本地事务。 4. TM 要求 TC 提交或回滚 XID 对应的全局事务。 5. TC 驱动 XID 对应的全局事务下的所有分支事务完成提交或回滚。 3 XAResource2.rollback Seata 柔性事务 整合 Seata AT 事务时，需要将 TM，RM 和 TC 的模型融入 Apache ShardingSphere 的分布式事务生态中。 在数据库资源上，Seata 通过对接 DataSource 接口，让 JDBC 操作可以同 TC 进行远程通信。同样，Apache ShardingSphere 也是面向 DataSource 接口，对用户配置的数据源进行聚合。因此，将

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

rollback 7.2.3 Seata 柔性事务 整合 Seata AT 事务时，需要将 TM，RM 和 TC 的模型融入 Apache ShardingSphere 的分布式事务生态中。 在数据库资源上，Seata 通过对接 DataSource 接口，让 JDBC 操作可以同 TC 进行远程通信。同样，Apache ShardingSphere 也是面向 DataSource 接口，对用户配置的数据源进行聚合。因此，将 开启全局事务 TM 控制全局事务的边界，TM 通过向 TC 发送 Begin 指令，获取全局事务 ID，所有分支事务通过此全局 事务 ID，参与到全局事务中；全局事务 ID 的上下文存放在当前线程变量中。 执行真实分片 SQL 处于 Seata 全局事务中的分片 SQL 通过 RM 生成 undo 快照，并且发送 participate 指令至 TC，加入 到全局事务中。由于 Apache ShardingSphere 的上下文传递。 7.2. 分布式事务 253 Apache ShardingSphere document, v5.0.0 提交或回滚事务 提交 Seata 事务时，TM 会向 TC 发送全局事务的提交或回滚指令，TC 根据全局事务 ID 协调所有分支事 务进行提交或回滚。 7.3 弹性伸缩 7.3.1 原理说明 考虑到 Apache ShardingSphere 的弹性伸缩模块的几个挑

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

transaction, we need to integrate TM, RM and TC component into Shard‐ ingSphere transaction manager. Seata have proxied DataSource interface in order to RPC with TC. Similarly, Apache ShardingSphere faced boundaries of global transactions. TM obtains the global transaction ID by sending Be‐ gin instructions to TC. All branch transactions participate in the global transaction through this global transaction ID transaction will be intercepted to generate undo snapshots by RM and sends participate instructions to TC to join global transaction. Since actual sharding SQLs executed in multi‐ threads, global transaction

the master node will be blocked during the re‐election process. e le ct io n: ra w‐ la te x: la tc h N O Distributed locks elected by the master nodeUsed for dis‐ tributed locks of curator s ha rd that need to failover from this node fa il ov er :r aw ‐l at ex :ˋ i te ms : ra w- la te x: `la tc h N O Distributed locks used when allocating failover shard items. Used by curator distributed