HIbernate "StaleStateException:批处理更新从update [0]返回意外行计数；实际行数: 0；预期: 1“

Question

当现有用户发出请求时，该方法首先删除最古老的记录，然后再保存新请求。如果请求不太快(使用Oracle DB)，下面的代码可以正常工作。

   public Request saveRequest(Request req)
   {
      String user = req.getUser();
      
      // Do a NamedQuery on the entity to pull all requests for user
      List<Request> requestList = getRequests(user);
      
      LOGGER.info("Request List size is " + requestList.size() + " for " + user);
      
      // Sort the list then delete the first/oldest request
      Comparator<Request> bySessionDate = Comparator.comparing(Request::getDate);
      Collections.sort(requestList, bySessionDate);
         
      LOGGER.info("Deleting request id " + requestList.get(0).getId());

      deleteById(requestList.get(0).getId());

      Request sreq = requestRepository.create(req);
      LOGGER.info("Saved request for user " + sreq.getUserId());
      return sreq;
   }

输出：

2021-01-20 00:39:12,167 INFO  [beez.service.RequestManager] (default task-4) Request List size is 250 for Bob
2021-01-20 00:39:12,168 INFO  [beez.service.RequestManager] (default task-4) Deleting request id 757A9B21E51D49199F2E182F68BC6BF7
2021-01-20 00:39:12,171 INFO  [beez.service.RequestManager] (default task-4) Deleted: 757A9B21E51D49199F2E182F68BC6BF7
2021-01-20 00:39:12,173 INFO  [beez.service.RequestManager] (default task-4) Saved request for user Bob
2021-01-20 00:39:15,375 INFO  [beez.service.RequestManager] (default task-3) Request List size is 250 for Bob
2021-01-20 00:39:15,375 INFO  [beez.service.RequestManager] (default task-3) Deleting request id 27239B85472C45EDA5495E98523295F3
2021-01-20 00:39:15,377 INFO  [beez.service.RequestManager] (default task-3) Deleted: 27239B85472C45EDA5495E98523295F3
2021-01-20 00:39:15,380 INFO  [beez.service.RequestManager] (default task-3) Saved request for user Bob

但是，如果用户以非常快的方式提交请求，即愉快的单击，那么相同的代码将生成一个StaleStateException错误。

2021-01-20 00:42:31,307 INFO  [beez.service.RequestManager] (default task-3) Request List size is 250 for Bob
2021-01-20 00:42:31,307 INFO  [beez.service.RequestManager] (default task-3) Deleting request id 55E43DF4D83E4BF5AD73DE47A49B0DA9
2021-01-20 00:42:31,310 INFO  [beez.service.RequestManager] (default task-3) Deleted: 55E43DF4D83E4BF5AD73DE47A49B0DA9
2021-01-20 00:42:31,313 INFO  [beez.service.RequestManager] (default task-3) Saved request for user Bob
2021-01-20 00:42:31,332 INFO  [beez.service.RequestManager] (default task-7) Request List size is 250 for Bob
2021-01-20 00:42:31,332 INFO  [beez.service.RequestManager] (default task-7) Deleting request id 55E43DF4D83E4BF5AD73DE47A49B0DA9
2021-01-20 00:42:31,492 ERROR [org.jboss.as.ejb3.invocation] (default task-7) WFLYEJB0034: EJB Invocation failed on component RequestManager for method public beez.entity.RequestManager service.RequestManager.saveRequest(beez.entity.RequestManager): javax.ejb.EJBTransactionRolledbackException: Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1
.
.
.
Caused by: org.hibernate.StaleStateException: Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1

似乎在下一个代码出现之前，代码还没有时间完成更改，导致代码尝试删除同一条记录两次。除了在此方法之前更改前端或其他东西之外，还有什么方法可以绕过吗？

我尝试过“事务处理”和“锁定”选项，但没有成功。在这个线程中花费了大量时间，但是这些解决方案要么不起作用，要么不适用：Hibernate - Batch update returned unexpected row count from update: 0 actual row count: 0 expected: 1

Answer 1

如果用户点击速度不快，强烈建议您考虑在前面删除API请求，以跳过请求/事件。

正如您所说的，由于加载到单个会话的实体与数据库的当前状态不一致，所以出现了正确的问题。

这可以通过多种方式解决，比如使用悲观锁定(select for update)和性能瓶颈，同步方法.等

处理此问题的最简单方法是使用JPA查询进行删除，并在数据库级别进行排序，因此这始终适用于记录表的当前状态。

delete from request where id= (select req.id from request req left join user usr on usr.id = req.userId where usr.userId=? order by req.date LIMIT 1)

请根据您的实体设计更正上述查询。