What is your experience with Spring Data (JPA, JDBC, Redis)? What interesting problems have you encountered with @Transactional?

Experience with Spring Data:

• JPA is used for ORM with Hibernate, convenient for working with relational databases, supports lazy loading, first-level caching.
• JDBC is applied for high-performance operations when fine-tuning SQL or batch processing is needed.
• Redis is integrated via Spring Data Redis for caching and session storage.

Problems with @Transactional:

• Proxies and internal method calls: if a method annotated with @Transactional is called directly from another method of the same class, a transaction is not created because Spring uses proxies. The solution is to move transactional methods to a separate bean or use AspectJ.

• Rollback on checked exceptions: by default, transactions rollback only on unchecked exceptions (RuntimeException). To rollback on checked exceptions, you need to explicitly specify rollbackFor.

• Non-standard propagation: sometimes it is challenging to choose the correct propagation (e.g., REQUIRES_NEW) to avoid issues with nested transactions.

Example:

@Service
public class UserService {

    @Transactional
    public void createUser(User user) {
        userRepository.save(user);
        // if a checked exception is thrown here, the transaction will not rollback without rollbackFor
    }

    @Transactional(rollbackFor = Exception.class)
    public void createUserWithRollback(User user) throws Exception {
        userRepository.save(user);
        throw new Exception("Force rollback");
    }
}

Overall, it is important to understand how Spring manages transactions and the limitations imposed by the proxy mechanism.