-
1. Re: Jboss 4.2.0CR1 -
kconner Mar 22, 2007 7:42 AM (in response to omerlin)You have not proved that there is a bug in TS.
What you have done is shown that if you start a new transaction then the currently aborted transaction is ignored.
The real question is *why* has the current transaction aborted. -
2. Re: Jboss 4.2.0CR1 -
kconner Mar 22, 2007 7:43 AM (in response to omerlin)You appended part of a server log file to the JCA thread but unfortunately this does not go back far enough.
Can you please send me the complete log file? -
3. Re: Jboss 4.2.0CR1 -
omerlin Mar 22, 2007 9:55 AM (in response to omerlin)Hello Kevin,
I sent to you directly the trace
Olivier -
4. Re: Jboss 4.2.0CR1 -
kconner Mar 22, 2007 10:38 AM (in response to omerlin)Hiya Olivier.
Nothing has arrived so far. -
5. Re: Jboss 4.2.0CR1 -
marklittle Mar 22, 2007 11:03 AM (in response to omerlin)Can you post the JCA cross reference here please?
-
7. Re: Jboss 4.2.0CR1 -
kconner Mar 22, 2007 7:04 PM (in response to omerlin)Hiya Olivier.
The log shows that the application has tried to enlist two LocalXAResources within the same transaction and the second one is causing the transaction to fail.
The Last Resource Commit Optimisation only succeeds with a single non XA Resource, adding more than one is transactionally unsafe. The version of TS which you are using aborts the transaction if this is attempted.
The solutions to this are as follows
- change your data source to be XA compliant
- run the two within different transactions (if it makes sense)
The next version of TS, coming soon, will allow multiple LocalXAResources to be enlisted within the same transaction. This functionality, as it is transactionally unsafe, will only be enabled if explicitly requested and will result in a warning message being issued everytime multiple LocalXAResources are enlisted.
Your best solution is to update your datasource. -
8. Re: Jboss 4.2.0CR1 -
weston.price Mar 22, 2007 9:33 PM (in response to omerlin)My, my...wonder where I heard of this before :-)
-
9. Re: Jboss 4.2.0CR1 -
omerlin Mar 23, 2007 5:14 AM (in response to omerlin)Hello,
thanks a lot for all these explanations - all seems clear now ...
Glassfish use by default XA datasource - that explains why it worked there and not on Jboss.
regards,
Olivier -
10. Re: Jboss 4.2.0CR1 -
marklittle Mar 23, 2007 6:26 AM (in response to omerlin)"weston.price@jboss.com" wrote:
My, my...wonder where I heard of this before :-)
Yes, but you'll notice that Glassfish has a proper datasource ;-) -
11. Re: Jboss 4.2.0CR1 -
weston.price Mar 23, 2007 6:41 AM (in response to omerlin)I'm sorry, do you work on Glassfish? You sound like your about twelve years old. Is this your first post? Do you need help in setting up an XA datasource?
Maybe you should read:
http://www.amazon.com/Java-Transaction-Processing-Implementation-Professional/dp/013035290X/ref=pd_bbs_sr_1/104-4885380-9366365?ie=UTF8&s=books&qid=1174646464&sr=8-1
Let me know how it works out for you! -
12. Re: Jboss 4.2.0CR1 -
omerlin Mar 23, 2007 6:53 AM (in response to omerlin)
Yes - the architectural choice was (not mine ) to develop on Glassfish (Full J2EE 5 ) implementation and to deploy on Jboss ... with its danger ...
As we are and i am not specialist on both we didn't notice the difference on default datasource configuration ... that could appear as an evidence after investigation and when you have experience on JTA.
Thanks to Kevin for its clear explanation
There no need to be polemical - simply be clear and factual like Kevin
rgds,
Olivier -
13. Re: Jboss 4.2.0CR1 -
weston.price Mar 23, 2007 6:54 AM (in response to omerlin)The comment was not directed towards you but Mark. I apologize for the confusion.
-
14. Re: Jboss 4.2.0CR1 -
weston.price Mar 23, 2007 6:55 AM (in response to omerlin)And it was tongue in cheek :-)