Issue
I have a problem where I am getting an: org.hibernate.MappingException: No Dialect mapping for JDBC type: 1111 when trying to call a postgres function using JPA create native query.
I created an EJB timer in a startup singleton to run a Postgres function every 6 hours. The function returns void and checks for expired records, deletes them, and updates some statuses. It takes no arguments and it returns void.
The postgres function runs perfectly if I call it using PgAdmin query tool (select function();) and returns void.
When I deploy the app on Glassfish 3.1.1 I get an exception and a failure to deploy.
This is the (shortened) stack trace:
WARNING: A system exception occurred during an invocation on EJB UserQueryBean method public void com.mysoftwareco.entity.utility.UserQueryBean.runRequestCleanup()
javax.ejb.TransactionRolledbackLocalException: Exception thrown from bean
...STACK TRACE BLAH BLAH BLAH ...
Caused by: javax.persistence.PersistenceException: org.hibernate.MappingException: No Dialect mapping for JDBC type: 1111
Here is the code:
First the JPA stuff:
public void runRequestCleanup() {
String queryString = "SELECT a_function_that_hibernate_chokes_on()";
Query query = em.createNativeQuery(queryString);
Object result = query.getSingleResult();
}
This is the singleton calling it:
@Startup
@Singleton
public class RequestCleanupTimer {
@Resource
TimerService timerService;
@EJB
UserQueryBean queryBean;
@PostConstruct
@Schedule(hour = "*/6")
void runCleanupTimer() {
queryBean.runRequestCleanup();
}
}
And the function:
CREATE OR REPLACE FUNCTION a_function_that_hibernate_chokes_on()
RETURNS void AS
$BODY$
DECLARE
var_field_id myTable.field_id%TYPE;
BEGIN
FOR var_field_id IN
select field_id from myTable
where status = 'some status'
and disposition = 'some disposition'
and valid_through < now()
LOOP
BEGIN
-- Do Stuff
END;
END LOOP;
END;
$BODY$
LANGUAGE plpgsql VOLATILE
COST 100;
Solution
I had enough messing around with JPA trying to get it to run a stored procedure.
I ended up using JDBC with a prepared statement. I did it in 15 minutes after spending several fruitless hours trying to fit a square peg into a round hole. I called the same jndi datasource my persistence unit uses to get a connection, created a prepared statement and closed it when done.
So if you need to run a stored procedure (or Postgres function) from a (now mostly) JPA app, here is what worked for me:
@Stateless
@LocalBean
public class UserQueryBean implements Serializable {
@Resource(mappedName="jdbc/DatabasePool")
private javax.sql.DataSource ds;
...
public void runRequestCleanup() {
String queryString = "SELECT cleanup_function_that_hibernateJPA_choked_on()";
Connection conn = null;
PreparedStatement statement = null;
try {
conn = ds.getConnection();
statement = conn.prepareCall(queryString);
statement.executeQuery();
} catch (SQLException ex) {
Logger.getLogger(UserQueryBean.class.getName()).log(Level.SEVERE, null, ex);
}finally{
try {
statement.close();
conn.close();
} catch (SQLException ex) {
Logger.getLogger(UserQueryBean.class.getName()).log(Level.SEVERE, null, ex);
}
}
// bit of logging code here
}
...
}
There seems to be a horrible oversight to leave out the simple ability to run a function or stored procedure on the server from JPA; especially one that doesn't return anything except void or the number of rows affected. And if it was deliberate ... no comment.
Edit: added close connection.
Answered By - Bill Rosmus
Answer Checked By - Cary Denson (JavaFixing Admin)