Hi :-),
the client does not stop automatically. 180secs after the commit was
called, there is a timeout fault on the server. Moreover I thought that
the coordinator would call either prepare or commit of my service. This
is not the case. Maybe I misunderstood the WS-AtomicTransaction
protocol. I thinkt the debit operation of the Web Service should only do
the work without further propagation to the underlying database. When
the commit operation of my Web Service is called, the operation
propagates to the database. Is that right?
Thanks,
Benjamin
Hi Benjamin,
> * completionCoordinator is called and it returns no response
> (containing <Commit xsi:nil="true"
> xmlns="http://schemas.xmlsoap.org/ws/2004/10/wsat"/>)
> * participant is called and it returns no response
> * coordinator is called and it returns no response
> * completionInitiator is called and it returns no response
That's OK, since those calls aren't expected to return anything. The
"Commit" message just tells the coordinator to commit. Once it has, it
sends a message to the completionInitiator containing the result -
Commited or Rolled Back.
(The same pattern applies to the calls between participant ("prepare")
and coordintor.)
The message exchange itself looks OK - please be more verbose on what
"fails" or what results you did not expect.
Just some general remarks - I don't think they apply to your case,
since the message exchange would fail earlier if something was not
correctly set up:
> I deployed all transactional services that are needed and put the
> client-config.wsdd to the WEB-INF/classes of my Tomcat.
Did you put the client-config.wsdd in both the classes directory of
the Tomcat that runs Kandula, *and* into the classpath of your client
program?
If something went wrong there, the messages lack the ws-addressing
fields that are necessary to correlate messages and transactions.
Please look in your messages if they contain To, From, and the like
fields.
Another guess: did you set the correct kandula location in both
kandula.properties files? (as above, there should be one in each
runtime instance. Of course, you could - or have - deployed everything
in one instance.)
Liebe Grüße ;-)
-hannes
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
