Using Narayana with Hibernate + Spring Boot

[mauriclaudio]

As you know, Spring Boot allows programmers to use Narayana, as well as other JTA implementatios (namely Atomikos and Bitronix), out of the box, by simply importing a starter dependency in your POM file and letting Spring Boot do the work for you.

As far as I understood, to use XA transactions you need to "wrap" your XADatasource in an helper class that is able to enlist a resource within an active transaction. For example, using Atomikos as JTA  manager implementation I need to instantiate a datasource this way

(i'm using MySQL as database server but that's not relevant):

public DataSource datasource(String dataSourceID, String url, String user, String pwd) {
     MysqlXADataSource mysqlXaDataSource = new MysqlXADataSource();
     mysqlXaDataSource.setUrl(url);
     mysqlXaDataSource.setPinGlobalTxToPhysicalConnection(true);
     mysqlXaDataSource.setPassword(pwd);
     mysqlXaDataSource.setUser(user);

     AtomikosDataSourceBean wrapper = new AtomikosDataSourceBean();
     wrapper.setUniqueResourceName(uniqueXAName);
     wrapper.setMaxPoolSize(MAX_POOL);
     wrapper.setMinPoolSize(MIN_POOL);
     wrapper.setXaDataSource(ds); 
     return wrapper;
}

Using Bitronix you have to use org.springframework.boot.jta.bitronix.PoolingDataSourceBean class instead of org.springframework.boot.jta.atomikos.AtomikosDataSourceBean, but class usage apart, you configure your wrapper the very same way, providing an unique id, setting pool size and so on.

With Narayama you have to use org.springframework.boot.jta.narayana.NarayanaDataSourceBean class, but it seems that with this wrapper you cannot set an unique id for your resource, nor set pool size in any way.

I wonder , first of all, if my assumption to use a "transaction manager aware" wrapper for any active datasource is or is not correct, and, in the case it's correct to use a wrapper, how to set pool size properly (or other properties, for example resource ID).

Can anyone help me ?

TIA

[zhfeng]

Hi mauriclaudio,

You are right to use the NarayanaDataSourceBean as the wrapper for the resource. And the Narayana transactional driver supports the pooling with [JBTM-2972] Add connection pooling facilities to transactionaldriver such that it will allow close to return the connect… , but I think it can not be set by the application.properties in the spring-boot currently. The other way is to use the common DBCP2 with the Narayana, we had a quickstart https://github.com/jbosstm/quickstart/tree/master/spring/narayana-spring-boot

and hope it is helpful.

Amos

[mauriclaudio]

Amos,

thanks a lot for your quick reply!  I will investigate further using DBCP2

Only as suggestion,  I think it would be nice if NarayanaDataSourceBean would allow programmer to  directly configure both datasource settings and other details, more or less what Atomikos or other provider allow users to do already.

Regards,

Claudio