Spring JPA ObjectOptimisticLockingFailureException，不使用批处理插入/更新

Question

为了为您提供有关我面临的问题的上下文，这是Postgres数据库中的客户表，其状态是EventHandler的更新，它从单个SQS队列中获取事件。 现在出现此错误

ObjectOptimisticLockingFailureException
Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1; statement executed: update customer set created_by=?, lock_id=?, modifiedat=?, modified_by=?, app_id=?, client=?, comments=?, customer_id=?, decision=?, source=? where id=? and lock_id=?; nested exception is org.hibernate.StaleStateException: Batch update returned unexpected row count from update [0]; actual row count: 0; expected: 1; statement executed: update customer set created_by=?, lock_id=?, modifiedat=?, modified_by=?, application_id=?, client_app=?, comments=?, customer_id=?, decision=?, source=? where id=? and lock_id=?

，此错误表明批处理更新正在发生，但是在此功能中，我正在执行批处理更新。 The code where this insert happens is

public Customer updateOrCreateCustomer(int customerId, String applicationId, String status) {
        Customer customer = customerRepository.findByCustomerId(customerId);

        if(customer == null) {
            customer = new Customer();
            customer.setCustomerId(customerId);
            customer.setApplicationId(applicationId);
            customer.status(status);
            log.info("Creating Customer with Customer Id - {} Application Id - {}", customerId, applicationId);
        } else {
            customer.setStatus(status);
            log.info("Updating Customer with Customer Id - {} Application Id - {}", customerId, applicationId);
        }
        return customerRepository.save(customer);
    }

Also, in my application.yml, I have set the batch_size property of JPA to 50 but this is being used in a different API where I需要做批处理插入物，

  jpa:
    hibernate:
      ddl-auto: none
    open-in-view: false
    properties:
      generate_statistics: false
      hibernate:
        order_inserts: true
        jdbc:
          batch_size: 50

我不知道为什么JPA在updateOrcreateCustomer函数中进行批处理更新。我猜是许多请求正在同一时间出现，因此JPA看到了batch_size已设置，因此它会自动将所有这些查询组合为一个，以优化插入/更新并做到这一点。请帮忙

Answer 1

我认为，如果您的对象在另一个线程中修改，然后尝试同时从另一个线程提交相同的对象，则可能会获得ObjectoptimisticlockingFailureException
我知道的解决方案是提高分离级别，以使项目一个一个一个合作，并且没有同步误差。

@Transactional(isolation = Isolation.ISOLATION_REPEATABLE_READ)