We Keep Coding

Why isn't my Hibernate insert reflected in my Hibernate query?

I've been asked to write some coded tests for a hibernate-based data access object.
I figure that I'd start with a trivial test: when I save a model, it should be in the collection returned by dao.getTheList(). The problem is, no matter what, when I call dao.getTheList(), it is always an empty collection.
The application code is already working in production, so let's assume that the problem is just with my test code.
#Test
#Transactional("myTransactionManager")
public void trivialTest() throws Exception {
...
// create the model to insert
...
session.save(model);
session.flush();
final Collection<Model> actual = dao.getTheList();
assertEquals(1, actual.size());
}
The test output is expected:<1> but was:<0>
So far, I've tried explicitly committing after the insert, and disabling the cache, but that hasn't worked.
I'm not looking to become a master of Hibernate, and I haven't been given enough time to read the entire documentation. Without really knowing where to start, this seemed like this might be a good question for the community.
What can I do to make sure that my Hibernate insert is flushed/committed/de-cached/or whatever it is, before the verification step of the test executes?
[edit] Some additional info on what I've tried. I tried manually committing the transaction between the insert and the call to dao.getTheList(), but I just get the error Could not roll back Hibernate transaction; nested exception is org.hibernate.TransactionException: Transaction not successfully started
#Test
#Transactional("myTransactionManager")
public void trivialTest() throws Exception {
...
// create the model to insert
...
final Transaction firstTransaction = session.beginTransaction();
session.save(model);
session.flush();
firstTransaction.commit();
final Transaction secondTransaction = session.beginTransaction();
final Collection<SystemConfiguration> actual = dao.getTheList();
secondTransaction.commit();
assertEquals(1, actual.size());
}
I've also tried breaking taking the #Transactional annotation off the test thread and annotating each of 2 helper methods, one for each Hibernate job. For that, though I get the error: No Hibernate Session bound to thread, and configuration does not allow creation of non-transactional one here.
[/edit]

I think the underlying DBMS might hide the change to other transactions as long as the changing transaction is not completed yet. Is getTheList running in an extra transaction? Are you using oracle or postgres?

Do I need to call JPA flush to check database constraints when my application is clustered?

I read that JPA caches SQL instructions to improve performance:
JPA providers like Hibernate can cache the SQL instructions they are
supposed to send to the database, often until you actually commit the
transaction. For example, you call em.persist(), Hibernate remembers
it has to make a database INSERT, but does not actually execute the
instruction until you commit the transaction.
I have a Java EE 6 application deployed to a Glassfish cluster with two instances. In the application there is a race condition where two Singletons do some expensive queries and then cache the results in a database table. They're doing the same work and trying to write the same record, so I sometimes get an exception:
java.sql.SQLIntegrityConstraintViolationException: ORA-00001: unique constraint (SOMESCHEMA.SOMETABLE_PK) violated
I decided the easiest way to deal with this would be to catch and ignore the exception:
// In a EJB with container-managed transactions.
public Entity getExpensiveEntity(int entityId) {
Entity entity = entityManager.find(Entity.class, entityId);
if (entity == null) {
try {
result = expensiveQueries();
entityManager.persist(result);
entityManager.flush();
} catch (SQLIntegrityConstraintViolationException ex) {
// The other instance already created the result, so get it.
result = jpa.find(result.getId());
}
}
return result;
}
I think the call to flush is necessary because otherwise the SQLIntegrityConstraintViolationException won't occur until the transaction ends somewhere up the EJB call stack, past catching and ignoring. Am I correct, is this a valid use case for flush? Is there a better way to handle this?
Reference
Correct use of flush() in JPA/Hibernate
A beginner’s guide to flush strategies in JPA and Hibernate
JPA and CMT -- Why Catching Persistence Exception is Not Enough?
JPA - create-if-not-exists entity? (see top-voted answer)

JPA exception handling

I'm running certain process that inserts new elements in DB from a for loop. I'm using JPA (Eclipselink), and sometimes there's a problem with the transation status. This is the case:
One of the INSERTS doesn't work (Primary Key duplicated)
After that, all the inserts will fail (Exception Description: Transaction is currently active).
for (Element l:e){
try{
//Should I add: if(!em.getTransaction().isActive())
em.getTransaction().begin();
em.createNativeQuery("INSERT INTO...").executeUpdate();
em.getTransaction().commit();
}
catch(Exception ep)
{
//right now I don't do anything here
}
}
I get that what is happening is that, since the commit in 1) didn't work, the transaction didn't finish, so the next em.getTransaction().begin() will find an already active transaction.
I have two ideas:
A) Before em.getTransaction().begin(), check if the transcation is active, and only if it is not, call begin(); otherwise, create query and commit.
B) Do something within the catch block. And here's my doubt... Should I call clear()? flush()? close()?
Which one looks better?
Thanks!

An exception thrown by ElementManager.Query does not rollback the active transaction. I see two options here:
Rollback the transaction by yourself within the catch clause with use of em.getTransaction().rollback().
Instead of inserting data with query use the preferred way based on EntityManager.persist whose exceptions cause an automatic rollback (in your particular case this will lead to javax.persistence.EntityExistsException).

Update database in the middle of the transaction, possible?

#Transactional
public void start() {
...
...
int result = entityManager
.createQuery("update Users set name=" + value + " where user.id=5").executeUpdate();
.....
}
Above code gives javax.persistence.TransactionRequiredException exception.
Update database in the middle of the transaction, possible ?
Any suggestions ?
Thanks.
A.
I just wonder if

This is a runtime exception which is thrown by the persistence provider when a transaction is required but is not active. A transaction is required because the start method is annotated as transactional. To get rid of the exception, you'll have to investigate why the line is called out of a transaction context.
A database update may be possible during a (different) transaction. Depends on the tables that are locked by the active transaction and on the transaction strategy. But in this case, it looks like you need to activate a transaction before you enter the start method.
With JPA you'd do something like this:
em = emf.createEntityManager();
tx = em.getTransaction();
tx.begin(); // now a transaction is active
start(); // call your method
// call other methods...
tx.commit(); // now the update is actually done
em.close();
Note - this is close to pseudo code, exception handling is missing in this example.

Nested transactions - Rollback scenario

A(){
con.begin;
.........
.........
B();
........
........(con.rollback;)
con.commit;
}
B{
con.begin;
.......
.......
con.commit;
}
In the above code, I begin a new DB transaction at A(). It executes some transaction successfully. After that B() starts executing and it also executes some transaction successfully and now the control returns to A(). At this point some exception occurs and I do a rollback. I would like to know whether the transaction which succeeded in B() will rollback or not.

The short answer, no. The long answer is as follows.
Support for nested transactions in Java depends on various variables at play.
Support for Nested transactions in JTA
First and foremost, if you are using JTA, it is upto to the Transaction Manager to support nested transactions. Any attempt to begin a transaction may result in a NotSupportedException being thrown by a Transaction Manager (that does not support nested transactions) if there is an attempt to start a new transaction in a thread that is already associated with a transaction.
From the Java Transaction API 1.1 specification:
3.2.1 Starting a Transaction
The TransactionManager.begin method starts
a global transaction and associates
the transaction context with the
calling thread. If the Transaction
Manager implementation does not
support nested transactions, the
TransactionManager.begin
methodthrowsthe NotSupportedException
whenthe calling thread is already
associated with a transaction.
Support for Nested transactions in JDBC
JDBC 3.0 introduces the Savepoint class, which is more or less similar to the concept of savepoints in the database. Savepoints have to be initialized using the Connection.setSavepoint() method that returns an instance of a Savepoint. One can roll back to this savepoint at a later point in time using the Connection.rollback(Savepoint svpt) method. All of this, of course, depends on whether you are using a JDBC 3.0 compliant driver that supports setting of savepoints and rolling back to them.
Impact of Auto-Commit
By default, all connections obtained are set to auto-commit, unless there is a clear deviation on this front by the JDBC driver. This feature, if enabled, automatically rules out the scope of having nested transactions, for all changes made in the database via the connection are committed automatically on execution.
If you disable the auto-commit feature, and choose to explicitly commit and rollback transactions, then committing a transaction always commits all changes performed by a connection until that point in time. Note, that the changes chosen for commit cannot be defined by a programmer - all changes until that instant are chosen for commit, whether they have been performed in one method or another. The only way out is to define savepoints, or hack your way past the JDBC driver - the driver usually commits all changes performed by a connection associated with a thread, so starting a new thread (this is bad) and obtaining a new connection in it, often gives you a new transaction context.
You might also want to check how your framework offers support for nested transactions, especially if you're isolated from the JDBC API or from starting new JTA transactions on your own.
Based on the above description of how nested transaction support is possibly achieved in various scenarios, it appears that a rollback in your code will rollback all changes associated with the Connection object.

That looks like poor transaction management i'm afraid. It would be good if you handle the commits and rollbacks from the callers to A and B instead.
A()
{
//business code A
B();
//more business code A
}
B()
{
//business code B
}
DoA()
{
try
{
con.begin();
A();
con.commit();
}
catch(Exception e)
{
con.rollback();
}
}
DoB()
{
try
{
con.begin();
B();
con.commit();
}
catch(Exception e)
{
con.rollback();
}
}

As per your code, in A() you are starting transaction. Then jump to B() where you start transaction again, which in turn will commit all previous transaction. Then at end of B(), transaction is explicitly committed. At this point, all your code is committed. Now the code return to A() and remaining code is processed. In case of exception, only this part after B() call will be rolled back.

You can use Java.SQL's built-in SavePoint function in Postgres 8 and up.
Connection conn = null;
Savepoint save = null;
DatabaseManager mgr = DatabaseManager.getInstance();
try {
conn = mgr.getConnection();
proc = conn.prepareCall("{ call writeStuff(?, ?) }");
//Set DB parameters
proc.setInt(1, stuffToSave);
proc.setString(2, moreStuff);
//Set savepoint here:
save = conn.setSavepoint();
//Try to execute the query
proc.execute();
//Release the savepoint, otherwise buffer memory will be eaten
conn.releaseSavepoint(save);
} catch (SQLException e) {
//You may want to log the first one only.
//This block will attempt to rollback
try {
//Rollback to the Savepoint of prior transaction:
conn.rollback(save);
} catch (SQLException e1) {
e1.printStackTrace();
}
}
When a SQL-exception occurs, the current transaction is rolled-back to the SavePoint, and the remaining transactions may occur. Without the roll-back, subsequent transactions will fail.