[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-04 Thread natebowler
I think I have found a fix. After looking through the source, I noticed another way to 
avoid that call was if isTrackByTx() is true.

So I added:

true

to my mssql-xa-ds.xml file. And, it appears to have addressed the issue. 

>From reading the docs, the only thing I can gather this does is hand control of 
>returning connections back to the pool over to the app server. And that the app 
>server will do this after the transaction is completed.

Is there anything else I should be aware of with this setting?


View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853815#3853815

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853815


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-03 Thread natebowler
BTW, 

The last thread I started in the TopLink support group can be found here. 

http://forums.oracle.com/forums/thread.jsp?forum=48&thread=273031&tstart=0&trange=30

Their initial response from the Oracle group is that a close() operation should not 
thrown an exception if already closed per the JTS spec.

Let me know if there is additional information I need to provide.

Nate

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853773#3853773

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853773


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-02 Thread natebowler
I don't believe there is still a currentTx in afterCompletion(). The incoming status 
to afterCompletion is 3 (STATUS_COMMITTED).

A  code snippet from the afterCompletion(int) handler that TopLink provides is as 
follows (hope this is helpful):

public void afterCompletion(int status) {
UnitOfWork unitOfWork = getUnitOfWork();
if (unitOfWork != null) {

if ( status == Status.STATUS_COMMITTED) {
 // THIS IS WHERE THE ERROR OCCURS!
// true=="committed"; true=="jts transaction"
unitOfWork.afterTransaction(true, true);

if(unitOfWork.isMergePending()) {
try {
unitOfWork.mergeClonesAfterCompletion();
getSession().decrementMergesPending();
} catch (RuntimeException rtEx) {
// Option to dump exception stack trace
if 
(shouldDumpExceptionsFromCallbacks()) {
rtEx.printStackTrace();

}
throw rtEx;
}
}
} else {
// false=="rolled back"; true=="jts transaction"
unitOfWork.afterTransaction(false, true);
}

unitOfWork.release();
((AbstractExternalTransactionController) 
getSession().getExternalTransactionController()).
removeActiveUnitOfWork(this.getGlobalTransaction());
}
}


As I mentioned before,the erroris thrown from JBoss after the TopLink call to 
UnitOfWork.afterTransaction(boolean, boolean). This is the stack trace:


org.jboss.resource.JBossResourceException: Error in delist!; - nested throwable: 
(java.lang.IllegalStateException: Already committed. 
TransactionImpl:XidImpl[FormatId=257, GlobalId=kungfu/13, BranchQual=, localId=13])
at 
org.jboss.resource.connectionmanager.BaseConnectionManager2.rethrowAsResourceException(BaseConnectionManager2.java:102)
at 
org.jboss.resource.connectionmanager.TxConnectionManager$TxConnectionEventListener.delist(TxConnectionManager.java:487)
at 
org.jboss.resource.connectionmanager.TxConnectionManager$TxConnectionEventListener.connectionClosed(TxConnectionManager.java:521)
at 
org.jboss.resource.adapter.jdbc.BaseWrapperManagedConnection.closeHandle(BaseWrapperManagedConnection.java:300)
at 
org.jboss.resource.adapter.jdbc.WrappedConnection.close(WrappedConnection.java:116)
at 
oracle.toplink.internal.databaseaccess.DatabaseAccessor.closeDatasourceConnection(DatabaseAccessor.java:326)
at 
oracle.toplink.internal.databaseaccess.DatasourceAccessor.closeConnection(DatasourceAccessor.java:346)
at 
oracle.toplink.internal.databaseaccess.DatasourceAccessor.afterJTSTransaction(DatasourceAccessor.java:73)
at oracle.toplink.threetier.ClientSession.afterTransaction(ClientSession.java:89)
at oracle.toplink.publicinterface.UnitOfWork.afterTransaction(UnitOfWork.java:1765)

This is close-source, but it appears all that is happening is the connection is simply 
being closed and returned to the connection pool.  After this operation, TopLink 
merges changes with the cache and continues cleanup.

If I remove this line of code, I begin to get a different exception in 
beginTransaction about Connections not being closed correctly..then connections run 
out. I'm not sure what options I have to make this work correctly.

My hope was that this is a case where the JTS spec is not clear and assumptions being 
made on the Toplink/JBoss side are in conflict.

Would more code samples and/or a simply binary/ear file help?

Nate

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853691#3853691

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853691


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-01 Thread [EMAIL PROTECTED]
I want to understand why you are getting the problem in
the first place.

The code change simply sidesteps the issue. It matches
the logic in JCA with the check done in JTA.
If JTA does not allow it, there is no point JCA trying it.

But why do you still have a currentTx when you try to close the
connection? What is the relative ordering of the afterCompletions
and what is the value of {is/was}trackByTx?

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853539#3853539

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853539


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-01 Thread natebowler
This is going to sound funny, but I think you've fixed it I made the change you 
outlined in the source and recompiled JBoss 4.0.0.

I don't get the error anymore. The whole test suite for my app just passed including 
the XA tests. Connections appear to be returning to the pool, etc.

Although the change may seem minor, it seems to directly address the issue I was 
having getting TopLink to work with JBoss. The transaction status was in conflict with 
TopLink's action to close the connection. 

I don't understand the JBoss source well enough to know what the overall impact is.  
I'm obviously going to spend more time with it, but it is very encouraging.

I think you're onto the problem, is there anything I can do to further test this 
issue? Any chance of this making it into the next JBoss 4.0.x version?

Nate

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853529#3853529

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853529


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-01 Thread [EMAIL PROTECTED]
The actual error message says it is trying to delist the XAResource
after it has been committed.
I don't see how this is possible unless there is some bizarre problem
with the ordering of the transaction synchronizations.
I did make some changes recently to avoid such a problem.
These were to remedy the problems you posted "ME TOO" on.
The fix should be in 3.2.6

One improvement I can think of by looking at the code is
TxConnectionManager.TxConnectionEventListener.delist() could have the following change:


  | - if (currentTx.getStatus() != Status.STATUS_NO_TRANSCATION)
  | +if (org.jboss.tm.TxUtils.isActive(currentTx))
  | 

Which will stop it trying to suspend the connection from the transaction
after the commit process has started.

This is more of a sanity check than a fix, since it looks like there
is some other ordering problem in the transaction synchronizations.

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853512#3853512

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853512


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-01 Thread [EMAIL PROTECTED]
Did you even "READ THIS FIRST"?

Posting "ME TOO" will just get you ignored. Unless your post
helps the original poster nobody is interested.

Sending private e-mails just makes you menace.

I recommend you "READ THIS FIRST" so you can see what
information we need to help you solve this problem.
A single stacktrace does not a bug report make.

How many times do I have post this response before it sinks in?

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853509#3853509

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853509


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user


[JBoss-user] [JCA/JBoss] - Re: TopLink (9.0.4-10.1.3) and JBoss (3.2.3-4.0.0) JTS XA in

2004-11-01 Thread [EMAIL PROTECTED]
We don't have toplink and don't support its usage so it would be better to create a 
bug report on sourceforge with a service that illustrates the interaction with the JTA 
interface toplink is attempting. If its integrating via the standard jca contract then 
we are interested in what the problem is. If its not, then its a question of the 
allowed operations from within the JTA callbacks.

http://sourceforge.net/tracker/?group_id=22866&atid=376685

View the original post : 
http://www.jboss.org/index.html?module=bb&op=viewtopic&p=3853500#3853500

Reply to the post : 
http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=3853500


---
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
___
JBoss-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-user