Thanks for all the options, I'll try that.

On 1/24/07, Marc Prud'hommeaux <[EMAIL PROTECTED]> wrote:

Matthieu-

My only other guess is that you might not actually be using your
ManagedRuntime setting, because we might need it to be a String (the
fully-qualified class name of your ManagedRuntime implementation),
rather than an actual instance of the class. Can you try specifying
the class name in the "openjpa.ManagedRuntime" setting? Also, try
enabling TRACE logging and posting the complete output from when you
run your test, since that might help show what the value is getting
set to.

If it's not that, then I'm afraid I'm stumped.

Can you try to debug it by either seeing if you can enable logging
for the JOTM TransactionManager, or by subclassing it? I want to see
if TransactionManager.registerSynchronization(broker) is ever being
called. Alternately, you can put in some debugging code in
AbstractBrokerFactory.syncWithManagedTransaction() and rebuilding
OpenJPA and seeing if that is ever called, although that might be
more work.



On Jan 24, 2007, at 2:49 PM, Matthieu Riou wrote:

> Any idea why my transaction manager setting seems to be ignored?
>
> Thanks,
> Matthieu
>
> On 1/16/07, Matthieu Riou <[EMAIL PROTECTED]> wrote:
>>
>> Hi,
>>
>> I've tried several thing:
>>
>> - First I made sure that the tx was started when I get the
>> EntityManager.
>> - Second I've tried calling em.joinTransaction() while the
>> transaction is
>> being executed, didn't change anything.
>> - Lastly I've tried to call em.flush() before commit and got the
>> following:
>>
>> Caused by: <4|false|0.9.7-incubating-SNAPSHOT>
>> org.apache.openjpa.persistence.TransactionRequiredException: Can only
>> perform operation while a transaction is active.
>>         at
>> org.apache.openjpa.kernel.BrokerImpl.assertActiveTransaction(
>> BrokerImpl.java:4250)
>>         at
>> org.apache.openjpa.kernel.DelegatingBroker.assertActiveTransaction(
>> DelegatingBroker.java:1292)
>>         at org.apache.openjpa.persistence.EntityManagerImpl.flush(
>> EntityManagerImpl.java:472)
>>         at org.apache.ode.bpel.engine.BpelDatabase$1.call(
>> BpelDatabase.java:79)
>>         at
>> org.apache.ode.bpel.scheduler.quartz.QuartzSchedulerImpl.execTransact
>> ion (
>> QuartzSchedulerImpl.java:245)
>>         at org.apache.ode.bpel.engine.BpelDatabase.exec
>> (BpelDatabase.java
>> :75)
>>
>> Seems that it doesn't see the transaction at all. Is there
>> something I
>> could check or make sure that the tx mgr is properly associated?
>>
>> Thanks,
>>
>> Matthieu
>>
>> On 1/15/07, Patrick Linskey <[EMAIL PROTECTED]> wrote:
>> >
>> > Also, what happens if you manually call flush() at the end of your
>> > persistence operations? This should at least tell us something
>> about
>> > whether or not the connections are being used correctly.
>> >
>> > -Patrick
>> >
>> > --
>> > Patrick Linskey
>> > BEA Systems, Inc.
>> >
>> >
>> _____________________________________________________________________
>> __
>> > Notice:  This email message, together with any attachments, may
>> contain
>> > information  of  BEA Systems,  Inc.,  its subsidiaries  and
>> affiliated
>> > entities,  that may be confidential,  proprietary,  copyrighted
>> and/or
>> > legally privileged, and is intended solely for the use of the
>> individual
>> > or entity named in this message. If you are not the intended
>> recipient,
>> > and have received this message in error, please immediately
>> return this
>> > by email and then delete it.
>> >
>> > > -----Original Message-----
>> > > From: Marc Prud'hommeaux [mailto: [EMAIL PROTECTED] On
>> > > Behalf Of Marc Prud'hommeaux
>> > > Sent: Monday, January 15, 2007 4:22 PM
>> > > To: open-jpa-dev@incubator.apache.org
>> > > Subject: Re: No tx commit when providing ManagedRuntime
>> > >
>> > > Matthieu-
>> > >
>> > > That's pretty weird. What happens if you just try to manually
>> commit
>> > > the transaction from the EntityManager itself (with
>> > > "em.getTransaction
>> > > ().commit()")?
>> > >
>> > > Also, has the global transaction already been started as the
>> point
>> > > when you obtain the EntityManager from the EntityManagerFactory?
>> > >
>> > > Finally, what happens if you call em.joinTransaction() and then
>> > > commit the global transaction. Any change?
>> > >
>> > >
>> > >
>> > > On Jan 15, 2007, at 3:41 PM, Matthieu Riou wrote:
>> > >
>> > > >> Does your TxMgrProvider provide a correctly functioning
>> > > >> TransactionManager? OpenJPA will register a
>> > > Synchronization with it,
>> > > >> which should get committed when your global transaction is
>> > > committed.
>> > > >
>> > > >
>> > > > It does, I directly provide JOTM transaction manager. The
>> > > same code
>> > > > works
>> > > > correctly with Hibernate as far as transaction handling is
>> > > > concerned. My
>> > > > managedRuntime impl is just:
>> > > >
>> > > >    public class TxMgrProvider implements ManagedRuntime {
>> > > >        public TxMgrProvider() {
>> > > >        }
>> > > >        public TransactionManager getTransactionManager() throws
>> > > > Exception {
>> > > >            return _txMgr;
>> > > >        }
>> > > >    }
>> > > >
>> > > > Do you see any log messages at all when you commit the global
>> > > >> transaction?
>> > > >
>> > > >
>> > > > No I don't. I only see some selects here and there. For example
>> > > > there should
>> > > > be a commit between the 2 following selects:
>> > > >
>> > > > 58194  ode-dao  TRACE  [Thread-4] openjpa.Query - Executing
>> query:
>> > > > select p
>> > > > from ProcessDAOImpl p
>> > > > 58205  ode-dao  TRACE  [Thread-4] openjpa.jdbc.SQL - <t
>> > > 21912919, conn
>> > > > 30031746> executing prepstmnt 8828117 SELECT t0.PROCESS_ID,
>> > > > t0.VERSION,
>> > > > t0.CONNECTION_ID , t0.GUID, t0.NUMBER_OF_INSTANCES,
>> t0.PROCESS_KEY,
>> > > > t0.PROCESS_TYPE FROM ODE_PROCESS t0
>> > > > 58206  ode-dao  TRACE  [Thread-4] openjpa.jdbc.SQL - <t
>> > > 21912919, conn
>> > > > 30031746> [1 ms] spent
>> > > > 58206  ode-dao  TRACE  [Thread-4] openjpa.jdbc.JDBC - <t
>> 21912919,
>> > > > conn
>> > > > 30031746> [0 ms] close
>> > > > 58208  ode-dao  TRACE  [Thread-4] openjpa.Query - Executing
>> query:
>> > > > select p
>> > > > from ProcessDAOImpl p
>> > > > 58208  ode-dao  TRACE  [Thread-4] openjpa.jdbc.SQL - <t
>> > > 21912919, conn
>> > > > 13257390> executing prepstmnt 15527192 SELECT t0.PROCESS_ID ,
>> > > > t0.VERSION,
>> > > > t0.CONNECTION_ID, t0.GUID, t0.NUMBER_OF_INSTANCES,
>> t0.PROCESS_KEY,
>> > > > t0.PROCESS_TYPE FROM ODE_PROCESS t0
>> > > > 58209  ode-dao  TRACE  [Thread-4] openjpa.jdbc.SQL - <t
>> > > 21912919, conn
>> > > > 13257390> [0 ms] spent
>> > > > 58209  ode-dao  TRACE  [Thread-4] openjpa.jdbc.JDBC - <t
>> 21912919,
>> > > > conn
>> > > > 13257390> [0 ms] close
>> > > >
>> > > > Thanks,
>> > > >
>> > > > Matthieu
>> > > >
>> > > > On Jan 15, 2007, at 10:35 AM, Matthieu Riou wrote:
>> > > >>
>> > > >> > Hi,
>> > > >> >
>> > > >> > I now have a working setup (at least something that
>> > > starts) using a
>> > > >> > programmaticaly provided datasource and transaction
>> manager. Here
>> > > >> > is the
>> > > >> > code now:
>> > > >> >
>> > > >> >        HashMap propMap = new HashMap();
>> > > >> >        propMap.put("openjpa.jdbc.DBDictionary", "
>> > > >> > org.apache.openjpa.jdbc.sql.DerbyDictionary");
>> > > >> >        propMap.put("openjpa.ManagedRuntime ", new
>> > > TxMgrProvider());
>> > > >> >        propMap.put("openjpa.ConnectionFactory",
>> _datasource);
>> > > >> >        propMap.put("openjpa.Log", "DefaultLevel=TRACE");
>> > > >> >        EntityManagerFactory emf =
>> > > >> > Persistence.createEntityManagerFactory("ode-dao",
>> > > >> > propMap);
>> > > >> >        HashMap propMap2 = new HashMap();
>> > > >> >        propMap2.put("openjpa.TransactionMode", "managed");
>> > > >> >        EntityManager em = emf.createEntityManager(propMap2);
>> > > >> >        _daoCF = new BPELDAOConnectionFactoryImpl(em);
>> > > >> >
>> > > >> > I've tried both with and without the propMap2, in both
>> cases I
>> > > >> > never get
>> > > >> > OpenJPA to commit (and therefore no insert/update/delete)
>> when I
>> > > >> > commit the
>> > > >> > transaction.
>> > > >> >
>> > > >> > Any idea of what could be wrong?
>> > > >> >
>> > > >> > Thanks!
>> > > >> >
>> > > >> > Matthieu
>> > > >>
>> > > >>
>> > >
>> > >
>> >
>>
>>


Reply via email to