Spring dynamic multiple data source configuration

Recently, many people have asked how to configure multiple data sources in spring, and it is troublesome to answer them one by one. Moreover, there are also problems with the configuration in previous blog posts, so I hereby re-release a Demo for everyone.

There are two data sources in the Demo, namely MySQL and Oracle. Simple tests have been carried out. There is no problem in dynamically switching data sources. I hope this Demo can help everyone.

In addition, I will give some instructions to explain the key points when configuring multiple data sources:

1. Pay attention to the configuration of the transaction interceptor

This is First and foremost. First of all, you need to understand that Spring's transaction management is bound to the data source. Once the program executes to the transaction management layer (such as service), since the transaction has been opened through the interceptor before entering this layer, so in this layer It is not possible to switch data sources. It is particularly important to understand the principles of transactions. In my previous article, there was a problem with configuring the interceptor for switching data sources in the Dao layer (because it was an example, so I was careless and misled me. I apologize to everyone), but the ideas provided are not problematic.

In Demo, the interceptor for switching data sources (dataSourceInterceptor) is configured on the upper layer of the transaction interceptor (txadvice), which is the Controller layer.

2. Pay attention to the creation of database tables

Some people like to use hibernate’s function of automatically creating tables, but it needs to be noted that in multiple data sources, especially different It is not possible to automatically create tables for all the multiple data sources in the database. Because Hibernate's automatic table creation is triggered when the project is started, only the tables of the default data source configured in the project will be created, and tables from other data sources will not be automatically created. Everyone should pay attention.

3. Hibernate’s database dialect (dialect) can be ignored

When there are multiple data sources, the dialect setting can be ignored. Hibernate will automatically recognize different ones when using it. Database, so you don’t have to worry about this configuration, or even not configure it.

4. Report No current session error

This is caused by using sessionFactory.getCurrentSession(). The current session is bound to the thread. One thread only A Session will be opened (no error will be reported unless openSession() is used), so the binding relationship between the session and the thread needs to be set.
Spring is used to manage Hibernate sessions in Demo, so OpenSessionInViewFilter is configured in web.xml and current_session_context_class is configured in hibernate.cfg.xml. [PS: When using Spring to manage Hibernate, you can remove hibernate.cfg.xml, and all configurations are in the Spring configuration file, that is, hibernateProperties. It depends on personal preference]

Finally, some frameworks are simply integrated into the Demo, which is convenient for everyone to use. It is a benefit to everyone. The configuration and technology of those frameworks will not be elaborated here.

The above is the entire content of this article. I hope it will be helpful to everyone's learning. I also hope that everyone will support the PHP Chinese website.

For more articles related to Spring dynamic multi-data source configuration, please pay attention to the PHP Chinese website!