使用悲观锁时发生Eclipse死锁

Question

第一个线程使用：

    Master master = entityManager.find(Master.class, "MASTER", LockModeType.PESSIMISTIC_WRITE);

据我理解，如果另一个线程试图执行相同的语句，它将挂起直到超时，然后抛出PessimisticLockingException或类似的语句。

然而，我的另一条主线似乎永远在尝试。我不明白为什么。在引导过程中，我清楚地看到了以下内容：

[EL Finest]: 2014-09-03 14:05:51.441--ServerSession(800990938)--Thread(Thread[main,5,main])--property=javax.persistence.lock.timeout; value=1

因此，它似乎应该配置为超时相当快，但它从来没有超时。然而，当第一个线程提交时，它立即继续，因此它的行为就像超时被设置为永远.

这是挂起的查询的日志：

[EL Finest]: 2014-09-03 14:05:52.124--UnitOfWork(189021365)--Thread(Thread[@Asynchronous classpath.ear - 1,5,main])--track pessimistic locked object com.kildeen.ref.domain.Master@42243b6 with UnitOfWork 189 021 365
//TABLE LOCKED BY FIRST THREAD
// lETS DO IT AGAIN IN ANOTHER THREAD
[EL Finer]: 2014-09-03 14:05:55.101--ServerSession(800990938)--Thread(Thread[main,5,main])--client acquired: 1749988512
[EL Finer]: 2014-09-03 14:05:55.102--UnitOfWork(430231018)--Thread(Thread[main,5,main])--TX binding to tx mgr, status=STATUS_ACTIVE
[EL Finer]: 2014-09-03 14:05:55.102--ClientSession(1749988512)--Thread(Thread[main,5,main])--acquire unit of work: 430231018
[EL Finest]: 2014-09-03 14:05:55.102--UnitOfWork(430231018)--Thread(Thread[main,5,main])--Execute query ReadObjectQuery(name="readObject" referenceClass=Master sql="SELECT ID, TESTPROPERTY FROM mp_master WHERE (ID = ?)")
[EL Finer]: 2014-09-03 14:05:55.102--ClientSession(1749988512)--Thread(Thread[main,5,main])--TX beginTransaction, status=STATUS_ACTIVE
[EL Finest]: 2014-09-03 14:05:55.102--ServerSession(800990938)--Connection(1844179384)--Thread(Thread[main,5,main])--Connection acquired from connection pool [default].
[EL Finest]: 2014-09-03 14:05:55.102--ClientSession(1749988512)--Thread(Thread[main,5,main])--reconnecting to external connection pool
[EL Fine]: 2014-09-03 14:05:55.102--ClientSession(1749988512)--Connection(1575043618)--Thread(Thread[main,5,main])--SELECT ID, TESTPROPERTY FROM mp_master WHERE (ID = ?) FOR UPDATE
    bind => [MASTER]

之后就陷入僵局..。

在Eclipselink 2.4.2中使用TomEE 1.7

Answer 1

切换到openjpa，它有更好的日志记录：

<openjpa-2.4.0-nonfinal-1598334-r422266:1599166 fatal general error> org.apache.openjpa.persistence.PersistenceException: The database dictionary in use ("class org.apache.openjpa.jdbc.sql.HSQLDictionary") reports that it does not have feature "SupportsSelectForUpdate".  This feature is needed to complete the current operation.  To force OpenJPA to try to use the feature anyway, set the following property:
openjpa.jdbc.DBDictionary: SupportsSelectForUpdate=<value>

结果发现它和postgres一起工作