Spring事务传播特性的浅析——事务方法嵌套调用的迷茫

http://www.iteye.com/topic/35907/  http://blog.csdn.net/hy6688_/article/details/44763869

Spring事务传播机制回顾 


   Spring事务一个被讹传很广说法是：一个事务方法不应该调用另一个事务方法，否则将产生两个事务。结果造成开发人员在设计事务方法时束手束脚，生怕一不小心就踩到地雷。 
其实这是不认识Spring事务传播机制而造成的误解，Spring对事务控制的支持统一在TransactionDefinition类中描述，该类有以下几个重要的接口方法： 

• int getPropagationBehavior()：事务的传播行为
• int getIsolationLevel()：事务的隔离级别
• int getTimeout()：事务的过期时间
• boolean isReadOnly()：事务的读写特性

   很明显，除了事务的传播行为外，事务的其他特性Spring是借助底层资源的功能来完成的，Spring无非只充当个代理的角色。但是事务的传播行为却是Spring凭借自身的框架提供的功能，是Spring提供给开发者最珍贵的礼物，讹传的说法玷污了Spring事务框架最美丽的光环。 
   
   所谓事务传播行为就是多个事务方法相互调用时，事务如何在这些方法间传播。Spring支持以下7种事务传播行为。 

• PROPAGATION_REQUIRED：如果当前没有事务，就新建一个事务，如果已经存在一个事务，就加入到这个事务中。这是最常见的选择。
• PROPAGATION_SUPPORTS：支持当前事务，如果当前没有事务，就以非事务方式执行。
• PROPAGATION_MANDATORY：使用当前的事务，如果当前没有事务，就抛出异常。
• PROPAGATION_REQUIRES_NEW：新建事务，如果当前存在事务，把当前事务挂起。
• PROPAGATION_NOT_SUPPORTED：以非事务方式执行操作，如果当前存在事务，就把当前事务挂起。
• PROPAGATION_NEVER：以非事务方式执行，如果当前存在事务，则抛出异常。
• PROPAGATION_NESTED：如果当前存在事务，则在嵌套事务内执行。如果当前没有事务，则执行与PROPAGATION_REQUIRED类似的操作。

   Spring默认的事务传播行为是PROPAGATION_REQUIRED，它适合绝大多数的情况，如果多个ServiveX#methodX()均工作在事务环境下（即均被Spring事务增强），且程序中存在如下的调用链：Service1#method1()->Service2#method2()->Service3#method3()，那么这3个服务类的3个方法通过Spring的事务传播机制都工作在同一个事务中。 

相互嵌套的服务方法  
   

   我们来看一下实例，UserService#logon()方法内部调用了UserService#updateLastLogon Time()和ScoreService#addScore()方法，这两个类都继承于BaseService。它们之间的类结构如下图所示：

   UserService#logon()方法内部调用了ScoreService#addScore()的方法，两者都分别通过Spring AOP进行了事务增强，则它们工作于同一事务中。来看具体的代码： 

 package com.baobaotao.nestcall;
 …
 @Service("userService")
 public class UserService extends BaseService {
     @Autowired
     private JdbcTemplate jdbcTemplate;

     @Autowired
     private ScoreService scoreService;

     //①该方法嵌套调用了本类的其他方法及其他服务类的方法
     public void logon(String userName) {
         System.out.println("before userService.updateLastLogonTime...");
         updateLastLogonTime(userName);//①-1本服务类的其他方法
         System.out.println("after userService.updateLastLogonTime...");

         System.out.println("before scoreService.addScore...");
         scoreService.addScore(userName, 20); //①-2其他服务类的其他方法
         System.out.println("after scoreService.addScore...");

     }
     public void updateLastLogonTime(String userName) {
         String sql = "UPDATE t_user u SET u.last_logon_time = ? WHERE user_name =?";
         jdbcTemplate.update(sql, System.currentTimeMillis(), userName);
     }

UserService中注入了ScoreService的Bean，而ScoreService的代码如下所示： 

 package com.baobaotao.nestcall;
 …
 @Service("scoreUserService")
 public class ScoreService extends BaseService{

     @Autowired
     private JdbcTemplate jdbcTemplate;

     public void addScore(String userName, int toAdd) {
         String sql = "UPDATE t_user u SET u.score = u.score + ? WHERE user_name =?";
         jdbcTemplate.update(sql, toAdd, userName);
     }
 }

  通过Spring配置为ScoreService及UserService中所有公有方法都添加Spring AOP的事务增强，让UserService的logon()和updateLastLogonTime()及ScoreService的addScore()方法都工作于事务环境下。下面是关键的配置代码： 

 "1.0" encoding="UTF-8" ?>
 "http://www.springframework.org/schema/beans"
        xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
        xmlns:context="http://www.springframework.org/schema/context"
        xmlns:p="http://www.springframework.org/schema/p" xmlns:aop="http://www.springframework.org/schema/aop"
        xmlns:tx="http://www.springframework.org/schema/tx"
        xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-3.0.xsd
     http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context-3.0.xsd http://www.springframework.org/schema/aop http://www.springframework.org/schema/aop/spring-aop-3.0.xsd http://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx-3.0.xsd">
     package="com.baobaotao.nestcall"/>
      …
     "jdbcManager"
           class="org.springframework.jdbc.datasource.DataSourceTransactionManager"
           p:dataSource-ref="dataSource"/>


     class="true">
         "serviceJdbcMethod"
                       expression="within(com.baobaotao.nestcall.BaseService+)"/>
         "serviceJdbcMethod" advice-ref="jdbcAdvice" order="0"/>

     "jdbcAdvice" transaction-manager="jdbcManager">

             "*"/>

   将日志级别设置为DEBUG，启动Spring容器并执行UserService#logon()的方法，仔细观察如下输出日志： 

 before userService.logon method...

      //①创建了一个事务
 Creating new transaction with name [com.baobaotao.nestcall.UserService.logon]: PROPAGATION_REQUIRED,ISOLATION_DEFAULT
 Acquired Connection [jdbc:mysql://localhost:3306/sampledb, UserName=root@localhost, MySQL-AB JDBC Driver] for JDBC transaction
 Switching JDBC Connection [jdbc:mysql://localhost:3306/sampledb, UserName=root@localhost, MySQL-AB JDBC Driver] to manual commit
 before userService.updateLastLogonTime...


 Executing prepared SQL update
 Executing prepared SQL statement [UPDATE t_user u SET u.last_logon_time = ? WHERE user_name =?]
 SQL update affected 1 rows
 after userService.updateLastLogonTime...
 before scoreService.addScore...

 //③ScoreService#addScore方法加入到①处启动的事务上下文中
 Participating in existing transaction
 Executing prepared SQL update
 Executing prepared SQL statement [UPDATE t_user u SET u.score = u.score + ? WHERE user_name =?]
 SQL update affected 1 rows
 after scoreService.addScore...
 Initiating transaction commit
 Committing JDBC transaction on Connection [jdbc:mysql://localhost:3306/sampledb, UserName=root@localhost, MySQL-AB JDBC Driver]
 …
 after userService.logon method...

    从上面的输出日志中，可以清楚地看到Spring为UserService#logon()方法启动了一个新的事务，而UserSerive#updateLastLogonTime()和UserService#logon()是在相同的类中，没有观察到有事务传播行为的发生，其代码块好像“直接合并”到UserService#logon()中。 
    然而在执行到ScoreService#addScore()方法时，我们就观察到发生一个事务传播的行为：" Participating in existing transaction "，这说明ScoreService#addScore()添加到UserService#logon()的事务上下文中，两者共享同一个事务。所以最终的结果是UserService的logon()、updateLastLogonTime()以及ScoreService的addScore都工作于同一事务中。 

注：以上内容摘自《Spring 3.x企业应用开发实战》 

解惑 spring 嵌套事务 

/** 
  * @author 王政 
  * @date 2006-11-24 
  * @note 转载请注明出处 
  */ 

   在所有使用 spring 的应用中, 声明式事务管理可能是使用率最高的功能了, 但是, 从我观察到的情况看, 
绝大多数人并不能深刻理解事务声明中不同事务传播属性配置的的含义, 让我们来看一下 TransactionDefinition 接口中的定义 

 /**
      * Support a current transaction, create a new one if none exists.
      * Analogous to EJB transaction attribute of the same name.
      * This is typically the default setting of a transaction definition.
      */
     int PROPAGATION_REQUIRED = 0;

     /**
      * Support a current transaction, execute non-transactionally if none exists.
      * Analogous to EJB transaction attribute of the same name.
      * Note: For transaction managers with transaction synchronization,
      * PROPAGATION_SUPPORTS is slightly different from no transaction at all,
      * as it defines a transaction scopp that synchronization will apply for.
      * As a consequence, the same resources (JDBC Connection, Hibernate Session, etc)
      * will be shared for the entire specified scope. Note that this depends on
      * the actual synchronization configuration of the transaction manager.
      * @see org.springframework.transaction.support.AbstractPlatformTransactionManager#setTransactionSynchronization
      */
     int PROPAGATION_SUPPORTS = 1;

     /**
      * Support a current transaction, throw an exception if none exists.
      * Analogous to EJB transaction attribute of the same name.
      */
     int PROPAGATION_MANDATORY = 2;

     /**
      * Create a new transaction, suspend the current transaction if one exists.
      * Analogous to EJB transaction attribute of the same name.
      * Note: Actual transaction suspension will not work on out-of-the-box
      * on all transaction managers. This in particular applies to JtaTransactionManager,
      * which requires the javax.transaction.TransactionManager to be
      * made available it to it (which is server-specific in standard J2EE).
      * @see org.springframework.transaction.jta.JtaTransactionManager#setTransactionManager
      */
     int PROPAGATION_REQUIRES_NEW = 3;

     /**
      * Execute non-transactionally, suspend the current transaction if one exists.
      * Analogous to EJB transaction attribute of the same name.
      * Note: Actual transaction suspension will not work on out-of-the-box
      * on all transaction managers. This in particular applies to JtaTransactionManager,
      * which requires the javax.transaction.TransactionManager to be
      * made available it to it (which is server-specific in standard J2EE).
      * @see org.springframework.transaction.jta.JtaTransactionManager#setTransactionManager
      */
     int PROPAGATION_NOT_SUPPORTED = 4;

     /**
      * Execute non-transactionally, throw an exception if a transaction exists.
      * Analogous to EJB transaction attribute of the same name.
      */
     int PROPAGATION_NEVER = 5;

     /**
      * Execute within a nested transaction if a current transaction exists,
      * behave like PROPAGATION_REQUIRED else. There is no analogous feature in EJB.
      * Note: Actual creation of a nested transaction will only work on specific
      * transaction managers. Out of the box, this only applies to the JDBC
      * DataSourceTransactionManager when working on a JDBC 3.0 driver.
      * Some JTA providers might support nested transactions as well.
      * @see org.springframework.jdbc.datasource.DataSourceTransactionManager
      */
     int PROPAGATION_NESTED = 6;

我们可以看到, 在 spring 中一共定义了六种事务传播属性, 如果你觉得看起来不够直观, 那么我来转贴一个满大街都有的翻译 

引用

PROPAGATION_REQUIRED -- 支持当前事务，如果当前没有事务，就新建一个事务。这是最常见的选择。 
PROPAGATION_SUPPORTS -- 支持当前事务，如果当前没有事务，就以非事务方式执行。 
PROPAGATION_MANDATORY -- 支持当前事务，如果当前没有事务，就抛出异常。 
PROPAGATION_REQUIRES_NEW -- 新建事务，如果当前存在事务，把当前事务挂起。 
PROPAGATION_NOT_SUPPORTED -- 以非事务方式执行操作，如果当前存在事务，就把当前事务挂起。 
PROPAGATION_NEVER -- 以非事务方式执行，如果当前存在事务，则抛出异常。 
PROPAGATION_NESTED -- 如果当前存在事务，则在嵌套事务内执行。如果当前没有事务，则进行与PROPAGATION_REQUIRED类似的操作。 
前六个策略类似于EJB CMT，第七个（PROPAGATION_NESTED）是Spring所提供的一个特殊变量。 
它要求事务管理器或者使用JDBC 3.0 Savepoint API提供嵌套事务行为（如Spring的DataSourceTransactionManager） 

在我所见过的误解中, 最常见的是下面这种: 

引用

假如有两个业务接口 ServiceA 和 ServiceB, 其中 ServiceA 中有一个方法实现如下 

/** 
* 事务属性配置为 PROPAGATION_REQUIRED 
*/ 
void methodA() { 
// 调用 ServiceB 的方法 
ServiceB.methodB(); 
} 

那么如果 ServiceB 的 methodB  如果配置了事务, 就必须配置为 PROPAGATION_NESTED 

这种想法可能害了不少人, 认为 Service 之间应该避免互相调用, 其实根本不用担心这点，PROPAGATION_REQUIRED 已经说得很明白, 
如果当前线程中已经存在事务, 方法调用会加入此事务, 果当前没有事务，就新建一个事务, 所以 ServiceB#methodB() 的事务只要遵循最普通的规则配置为 PROPAGATION_REQUIRED 即可, 如果 ServiceB#methodB (我们称之为内部事务, 为下文打下基础) 抛了异常, 那么 ServiceA#methodA(我们称之为外部事务) 如果没有特殊配置此异常时事务提交 (即 +MyCheckedException的用法), 那么整个事务是一定要 rollback 的, 什么 Service 只能调 Dao 之类的言论纯属无稽之谈, spring 只负责配置了事务属性方法的拦截, 它怎么知道你这个方法是在 Service 还是 Dao 里 ? 

     说了这么半天, 那到底什么是真正的事务嵌套呢, 解释之前我们来看一下  Juergen Hoeller 的原话 

Juergen Hoeller 写道

PROPAGATION_REQUIRES_NEW starts a new, independent "inner" transaction for the given scope. This transaction will be committed or rolled back completely independent from the outer transaction, having its own isolation scope, its own set of locks, etc. The outer transaction will get suspended at the beginning of the inner one, and resumed once the inner one has completed. 

Such independent inner transactions are for example used for id generation through manual sequences, where the access to the sequence table should happen in its own transactions, to keep the lock there as short as possible. The goal there is to avoid tying the sequence locks to the (potentially much longer running) outer transaction, with the sequence lock not getting released before completion of the outer transaction. 

PROPAGATION_NESTED on the other hand starts a "nested" transaction, which is a true subtransaction of the existing one. What will happen is that a savepoint will be taken at the start of the nested transaction. íf the nested transaction fails, we will roll back to that savepoint. The nested transaction is part of of the outer transaction, so it will only be committed at the end of of the outer transaction. 

Nested transactions essentially allow to try some execution subpaths as subtransactions: rolling back to the state at the beginning of the failed subpath, continuing with another subpath or with the main execution path there - all within one isolated transaction, and not losing any previous work done within the outer transaction. 

For example, consider parsing a very large input file consisting of account transfer blocks: The entire file should essentially be parsed within one transaction, with one single commit at the end. But if a block fails, its transfers need to be rolled back, writing a failure marker somewhere. You could either start over the entire transaction every time a block fails, remembering which blocks to skip - or you mark each block as a nested transaction, only rolling back that specific set of operations, keeping the previous work of the outer transaction. The latter is of course much more efficient, in particular when a block at the end of the file fails. 

Juergen Hoeller 写道

Rolling back the entire transaction is the choice of the demarcation code/config that started the outer transaction. 

So if an inner transaction throws an exception and is supposed to be rolled back (according to the rollback rules), the transaction will get rolled back to the savepoint taken at the start of the inner transaction. The immediate calling code can then decide to catch the exception and proceed down some other path within the outer transaction. 

If the code that called the inner transaction lets the exception propagate up the call chain, the exception will eventually reach the demarcation code of the outer transaction. At that point, the rollback rules of the outer transaction decide whether to trigger a rollback. That would be a rollback of the entire outer transaction then. 

So essentially, it depends on your exception handling. If you catch the exception thrown by the inner transaction, you can proceed down some other path within the outer transaction. If you let the exception propagate up the call chain, it's eventually gonna cause a rollback of the entire outer transaction. 

    也就是说, 最容易弄混淆的其实是 PROPAGATION_REQUIRES_NEW 和 PROPAGATION_NESTED, 那么这两种方式又有何区别呢? 我简单的翻译一下 Juergen Hoeller 的话 : 
    
     PROPAGATION_REQUIRES_NEW 启动一个新的, 不依赖于环境的 "内部" 事务. 这个事务将被完全 commited 或 rolled back 而不依赖于外部事务, 它拥有自己的隔离范围, 自己的锁, 等等. 当内部事务开始执行时, 外部事务将被挂起, 内务事务结束时, 外部事务将继续执行.  


     另一方面, PROPAGATION_NESTED 开始一个 "嵌套的" 事务,  它是已经存在事务的一个真正的子事务. 潜套事务开始执行时,  它将取得一个 savepoint. 如果这个嵌套事务失败, 我们将回滚到此 savepoint. 潜套事务是外部事务的一部分, 只有外部事务结束后它才会被提交.  

     由此可见, PROPAGATION_REQUIRES_NEW 和 PROPAGATION_NESTED 的最大区别在于, PROPAGATION_REQUIRES_NEW 完全是一个新的事务, 而 PROPAGATION_NESTED 则是外部事务的子事务, 如果外部事务 commit, 潜套事务也会被 commit, 这个规则同样适用于 roll back.  
    
    
    那么外部事务如何利用嵌套事务的 savepoint 特性呢, 我们用代码来说话 
    

 ServiceA {

     /**
      * 事务属性配置为 PROPAGATION_REQUIRED
      */
     void methodA() {
         ServiceB.methodB();
     }

 }

 ServiceB {

     /**
      * 事务属性配置为 PROPAGATION_REQUIRES_NEW
      */
     void methodB() {
     }

 }

    

这种情况下, 因为 ServiceB#methodB 的事务属性为 PROPAGATION_REQUIRES_NEW, 所以两者不会发生任何关系, ServiceA#methodA 和 ServiceB#methodB 不会因为对方的执行情况而影响事务的结果, 因为它们根本就是两个事务, 在 ServiceB#methodB 执行时 ServiceA#methodA 的事务已经挂起了 (关于事务挂起的内容已经超出了本文的讨论范围, 有时间我会再写一些挂起的文章) . 

那么 PROPAGATION_NESTED 又是怎么回事呢? 继续看代码 

 ServiceA {

     /**
      * 事务属性配置为 PROPAGATION_REQUIRED
      */
     void methodA() {
         ServiceB.methodB();
     }

 }

 ServiceB {

     /**
      * 事务属性配置为 PROPAGATION_NESTED
      */
     void methodB() {
     }

 }

现在的情况就变得比较复杂了, ServiceB#methodB 的事务属性被配置为 PROPAGATION_NESTED, 此时两者之间又将如何协作呢? 从 Juergen Hoeller 的原话中我们可以找到答案, ServiceB#methodB 如果 rollback, 那么内部事务(即 ServiceB#methodB) 将回滚到它执行前的 SavePoint(注意, 这是本文中第一次提到它, 潜套事务中最核心的概念), 而外部事务(即 ServiceA#methodA) 可以有以下两种处理方式: 

1. 改写 ServiceA 如下 

 ServiceA {

     /**
      * 事务属性配置为 PROPAGATION_REQUIRED
      */
     void methodA() {
         try {
             ServiceB.methodB();
         } catch (SomeException) {
             // 执行其他业务, 如 ServiceC.methodC();
         }
     }

 }

这种方式也是潜套事务最有价值的地方, 它起到了分支执行的效果, 如果 ServiceB.methodB 失败, 那么执行 ServiceC.methodC(), 而 ServiceB.methodB 已经回滚到它执行之前的 SavePoint, 所以不会产生脏数据(相当于此方法从未执行过), 这种特性可以用在某些特殊的业务中, 而 PROPAGATION_REQUIRED 和 PROPAGATION_REQUIRES_NEW 都没有办法做到这一点. (题外话 : 看到这种代码, 似乎似曾相识, 想起了 prototype.js 中的 Try 函数 ) 

2. 代码不做任何修改, 那么如果内部事务(即 ServiceB#methodB) rollback, 那么首先 ServiceB.methodB 回滚到它执行之前的 SavePoint(在任何情况下都会如此), 
   外部事务(即 ServiceA#methodA) 将根据具体的配置决定自己是 commit 还是 rollback (+MyCheckedException). 
   
   
上面大致讲述了潜套事务的使用场景, 下面我们来看如何在 spring 中使用 PROPAGATION_NESTED, 首先来看 AbstractPlatformTransactionManager 

 /**
  * Create a TransactionStatus for an existing transaction.
  */
 private TransactionStatus handleExistingTransaction(
         TransactionDefinition definition, Object transaction, boolean debugEnabled)
         throws TransactionException {

    ... 省略

     if (definition.getPropagationBehavior() == TransactionDefinition.PROPAGATION_NESTED) {
         if (!isNestedTransactionAllowed()) {
             throw new NestedTransactionNotSupportedException(
                     "Transaction manager does not allow nested transactions by default - " +
                     "specify 'nestedTransactionAllowed' property with value 'true'");
         }
         if (debugEnabled) {
             logger.debug("Creating nested transaction with name [" + definition.getName() + "]");
         }
         if (useSavepointForNestedTransaction()) {
             // Create savepoint within existing Spring-managed transaction,
             // through the SavepointManager API implemented by TransactionStatus.
             // Usually uses JDBC 3.0 savepoints. Never activates Spring synchronization.
             DefaultTransactionStatus status =
                     newTransactionStatus(definition, transaction, false, false, debugEnabled, null);
             status.createAndHoldSavepoint();
             return status;
         }
         else {
             // Nested transaction through nested begin and commit/rollback calls.
             // Usually only for JTA: Spring synchronization might get activated here
             // in case of a pre-existing JTA transaction.
             doBegin(transaction, definition);
             boolean newSynchronization = (this.transactionSynchronization != SYNCHRONIZATION_NEVER);
             return newTransactionStatus(definition, transaction, true, newSynchronization, debugEnabled, null);
         }
     }
 }

   
一目了然 

1. 我们要设置 transactionManager 的 nestedTransactionAllowed 属性为 true, 注意, 此属性默认为 false!!!  

再看 AbstractTransactionStatus#createAndHoldSavepoint() 方法 

 /**
  * Create a savepoint and hold it for the transaction.
  * @throws org.springframework.transaction.NestedTransactionNotSupportedException
  * if the underlying transaction does not support savepoints
  */
 public void createAndHoldSavepoint() throws TransactionException {
     setSavepoint(getSavepointManager().createSavepoint());
 }

  可以看到 Savepoint 是 SavepointManager.createSavepoint 实现的, 再看 SavepointManager 的层次结构, 发现 
  其 Template 实现是 JdbcTransactionObjectSupport, 常用的 DatasourceTransactionManager, HibernateTransactionManager
  中的 TransactonObject 都是它的子类 : 

 
   
  JdbcTransactionObjectSupport 告诉我们必须要满足两个条件才能 createSavepoint : 
  
2. java.sql.Savepoint 必须存在, 即 jdk 版本要 1.4+ 
3. Connection.getMetaData().supportsSavepoints() 必须为 true, 即 jdbc drive 必须支持 JDBC 3.0 

确保以上条件都满足后, 你就可以尝试使用 PROPAGATION_NESTED 了. (全文完)