事务的传播类型介绍
REQUIRED:支持当前事务,如果当前不存在则新开启一个事务(默认配置)
SUPPORTS:支持当前事务,如果当前不存在事务则以非事务方式执行
MANDATORY:支持当前事务,如果当前不存在事务则抛出异常
REQUIRES_NEW:创建一个新事务,如果当前已存在事务则挂起当前事务
NOT_SUPPORTED:以非事务方式执行,如果当前已存在事务则挂起当前事务
NEVER:以非事务方式执行,如果当前已存在事务则抛出异常
NESTED:如果当前存在事务,则在嵌套事务中执行,否则开启一个新事务
案例演示数据说明
数据表
CREATE TABLE `t1` (
`id` int(11) NOT NULL,
`c` int(11) DEFAULT NULL,
PRIMARY KEY (`id`),
KEY `idx_c` (`c`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
CREATE TABLE `t2` (
`id` int(11) NOT NULL,
`c` int(11) DEFAULT NULL,
PRIMARY KEY (`id`),
KEY `idx` (`c`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
<update id="updateT1">
update t1 set c = 2 where id = 1
</update>
<update id="updateT2">
update t2 set c = 2 where id = 1
</update>
REQUIRED
当调用T1Service中的func方法时,除了更新t1表数据外,还会调用t2Service的func方法,更新t2表。
@Service
public class T1Service {
@Resource
private TestMapper testMapper;
@Resource
private T2Service t2Service;
@Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
}
}
@Service
public class T2Service {
@Resource
private TestMapper testMapper;
@Transactional
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
}
@Transactional默认的传播方式就是REQUIRED,所以当方法执行到int i = 1 / 0时,会抛出异常,t1、t2表中的数据都不会被修改。
SUPPORTS
@Service
public class T1Service {
@Resource
private TestMapper testMapper;
@Resource
private T2Service t2Service;
// @Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
}
}
@Service
public class T2Service {
@Resource
private TestMapper testMapper;
/**
* 数据不会回滚,因为当前没有事务,SUPPORTS会以非事务方式执行
*/
@Transactional(propagation = Propagation.SUPPORTS)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
}
MANDATORY
当t1Service没有事务时,把t2Service的func方法,配置为@Transactional(propagation = Propagation.MANDATORY)
// t1Service
public void func() {
testMapper.updateT1();
t2Service.func();
}
// t2Service
@Transactional(propagation = Propagation.MANDATORY)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
抛出异常
org.springframework.transaction.IllegalTransactionStateException: No existing transaction found for transaction marked with propagation 'mandatory'
at org.springframework.transaction.support.AbstractPlatformTransactionManager.getTransaction(AbstractPlatformTransactionManager.java:362) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionAspectSupport.createTransactionIfNecessary(TransactionAspectSupport.java:595) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionAspectSupport.invokeWithinTransaction(TransactionAspectSupport.java:382) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionInterceptor.invoke(TransactionInterceptor.java:119) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:186) ~[spring-aop-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.CglibAopProxy$CglibMethodInvocation.proceed(CglibAopProxy.java:753) ~[spring-aop-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.CglibAopProxy$DynamicAdvisedInterceptor.intercept(CglibAopProxy.java:698) ~[spring-aop-5.3.14.jar:5.3.14]
REQUIRES_NEW
毫无疑问,t2的数据不会被更新,当没有事务时,REQUIRES_NEW会自己创建一个事务
// t1Service
public void func() {
testMapper.updateT1();
t2Service.func();
}
// t2Service
@Transactional(propagation = Propagation.REQUIRES_NEW)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
与REQUIRED有什么区别呢?
现在把抛出异常的地方放到t1Service中
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
int i = 1 / 0;
}
// t2Service
@Transactional(propagation = Propagation.REQUIRES_NEW)
public void func() {
testMapper.updateT2();
}
此时执行后,t2的数据不会回滚,t1的数据会回滚,因为t2和t1不是一个事务。
NOT_SUPPORTED
NOT_SUPPORTED的效果就是无论异常是在t1Service还是t2Service都不会回滚t2的数据。
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
int i = 1 / 0;
}
// t2Service
@Transactional(propagation = Propagation.NOT_SUPPORTED)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
NEVER
很明显,如果存在事务,直接抛出异常
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
}
// t2Service
@Transactional(propagation = Propagation.NEVER)
public void func() {
testMapper.updateT2();
}
org.springframework.transaction.IllegalTransactionStateException: Existing transaction found for transaction marked with propagation 'never'
at org.springframework.transaction.support.AbstractPlatformTransactionManager.handleExistingTransaction(AbstractPlatformTransactionManager.java:413) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.support.AbstractPlatformTransactionManager.getTransaction(AbstractPlatformTransactionManager.java:352) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionAspectSupport.createTransactionIfNecessary(TransactionAspectSupport.java:595) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionAspectSupport.invokeWithinTransaction(TransactionAspectSupport.java:382) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.transaction.interceptor.TransactionInterceptor.invoke(TransactionInterceptor.java:119) ~[spring-tx-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:186) ~[spring-aop-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.CglibAopProxy$CglibMethodInvocation.proceed(CglibAopProxy.java:753) ~[spring-aop-5.3.14.jar:5.3.14]
at org.springframework.aop.framework.CglibAopProxy$DynamicAdvisedInterceptor.intercept(CglibAopProxy.java:698) ~[spring-aop-5.3.14.jar:5.3.14]
如果把t1Service中的事务去掉,则没问题,但t2Service抛出异常后,也不会回滚
// t1Service
public void func() {
testMapper.updateT1();
t2Service.func();
}
// t2Service
@Transactional(propagation = Propagation.NEVER)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
NESTED
NESTED应该是几种事务传播方式中最难理解的,如果不注意,NESTED和REQUIRED功能看起来则差不多,都可以理解为有事务则加入,没有则新启一个,但实际上NESTED比REQUIRED要更加灵活。
先来看第一个案例,在t1Service中调用t2Service时,对异常进行了捕获,并且也没有抛出。
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
try {
t2Service.func();
} catch (Exception e) {
e.printStackTrace();
}
}
// t2Service
@Transactional(propagation = Propagation.REQUIRED)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
当t2Service配置为REQUIRED时,t1、t2都进行了回滚,因为是同一个事务,当如果t2Service配置为NESTED就不一样了,此时t1则不会回滚。
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
try {
t2Service.func();
} catch (Exception e) {
e.printStackTrace();
}
}
// t2Service
@Transactional(propagation = Propagation.NESTED)
public void func() {
testMapper.updateT2();
int i = 1 / 0;
}
NESTED与REQUIRES_NEW的区别
现在有人可能觉得NESTED和REQUIRES_NEW有点相似,但实际上要注意NESTED和REQUIRES_NEW是很有大的区别的
现在我们分别给t1Service和t2Service加上一个TransactionSynchronizationManager.getCurrentTransactionName()输出看看效果
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
System.out.println(TransactionSynchronizationManager.getCurrentTransactionName());
try {
t2Service.func();
} catch (Exception e) {
e.printStackTrace();
}
}
// t2Service
@Transactional(propagation = Propagation.REQUIRES_NEW)
public void func() {
testMapper.updateT2();
System.out.println(TransactionSynchronizationManager.getCurrentTransactionName());
int i = 1 / 0;
}
输出结果
com.demo.transaction.service.T1Service.func
com.demo.transaction.service.T2Service.func
把REQUIRES_NEW替换为NESTED,可以看出使用NESTED后,实际上还是同一个事务。
com.demo.transaction.service.T1Service.func
com.demo.transaction.service.T1Service.func
NESTED实现方式
这就是NESTED不同之处,两个方法同一个事务,居然没有一起回滚,这就叫嵌套事务,子事务回滚不会影响到主事务,实际上利用的是savepoint功能,就好像下面这样
-- 主事务
savepoint;
-- 执行主事务代码
-- 子事务
savepoint;
-- 执行子事务代码
-- 子事务提交
commit;
-- 执行主事务代码
-- 主事务提交
commit;
所以,如果是在主事务中抛出异常,那么子事务也会被回滚,就像下面这样。
// t1Service
@Transactional
public void func() {
testMapper.updateT1();
t2Service.func();
int i = 1 / 0;
}
// t2Service
@Transactional(propagation = Propagation.NESTED)
public void func() {
testMapper.updateT2();
}