Bugs item #792120, was opened at 2003-08-20 19:21
Message generated for change (Comment added) made by d_jencks
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=376685&aid=792120&group_id=22866

Category: JBossCX
Group: v3.2
Status: Open
Resolution: None
Priority: 5
Submitted By: Jim Cook (oravecz)
Assigned to: Nobody/Anonymous (nobody)
Summary: Local transaction broken when using pass-by-value

Initial Comment:
jBoss 3.2.2 branch, HEAD

In order to get transactional support when using local-
tx-datasource, each time a connection is fetched from 
the pool, the server is supposed to hand you the same 
underlying connection object as long as you are in the 
same transactional context (and on the same server, of 
course). I assume that the server uses some sort of 
ThreadLocal-type algorithm to keep this straight.

Now, since switching to a pass-by-value configuration, 
it seems that the server is no longer handing back the 
same underlying connection. In fact, in a single 
J2EE "transaction" is creating 3 connections to the 
database. Since this is not an XA transactional setup, 
these connections are not actually transactional. They 
are also causing all kinds of deadlocks also, but that 
would be expected given the problem.

I have attached a log4j output that demonstrates that 
a new connection to Sybase is created for every RMI 
thread in use. Since we are using pass-by-value JNDI, a 
new RMI thread is created every time we leave a 
deployment context. I had to move to pass-by-value to 
satisfy the class loading restrictions imposed by jBoss 
3.2.


----------------------------------------------------------------------

Comment By: David Jencks (d_jencks)
Date: 2003-08-22 03:34

Message:
Logged In: YES 
user_id=60525

The local tx connection manager uses a tx to ManagedConnection 
map to keep track of the transaction to ManagedConnection 
association.  I think it is unlikely that this is failing.  

If anyone investigates this, I would suggest starting by 
determining if the new "RMI thread" actually has the correct 
transaction associated with it.  I don't know of any mechanism in 
JBoss 3.2 by which this could occur.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=376685&aid=792120&group_id=22866


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0
_______________________________________________
JBoss-Development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development

Reply via email to