activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Li, Fan" <fa...@amazon.com>
Subject RE: PrefetchSubscription preLoadSize not decreased
Date Wed, 22 Mar 2006 19:28:23 GMT
Hi:

After getting the new code, the problem is fixed.

Thanks
Fan

-----Original Message-----
From: chirino@gmail.com [mailto:chirino@gmail.com] On Behalf Of Hiram Chirino
Sent: Wednesday, March 22, 2006 10:18 AM
To: activemq-dev@geronimo.apache.org
Subject: Re: PrefetchSubscription preLoadSize not decreased

Hi Li,

As of svn revision 387562, we no longer use the preLoadSize.  see:
http://svn.apache.org/viewcvs?rev=387562&view=rev

Please get an update and let me know if this has fixed your issue.

On 3/22/06, Li, Fan <fanli@amazon.com> wrote:
> Hi:
>
> I am running test on the JmsConnectors, the test case I am trying to run is sending and
receiving Messages in the Queue region and the transport I am using is TcpTransport.
>
> I am running into a problem with the preLoadSize of PrefetchSubscription not being decremented
at all, so Messages add to PrefetchSubscription can no longer be dispatched. The preLoadSize
is decreased by calling the decrementPreloadSize of PrefetchSubscription, which is never called
in my test case.
>
> I have setup a similar test but using VMTransport, there appears to be no problem. In
both test cases, the dispatch method of PrefetchSubscription calls the dispatchAsync method
of its associated AbstractConnection for dispatching. The differences between the two test
cases I have notices is that when the transport is tcp, there is a taskRunner associated with
the AbstactConnection, which will be wakeup to run the task at some future time; but when
the transport is vm, the AbstractConnection has no taskRunner and the dispatchSync method
will be called to dispatch the message as well as calling the  run method on the Consumer
of the message, which will correctly call the onDispatch method of PrefetchSubscription and
decrease the preLoadSize.
>
> I have trouble tracing whether or not the task that eventually ran by the task associated
with the taskRunner actually called the run method on the Consumer of the message, do anyone
knows how to fix this problem?
>
> Thanks
> Fan
>
>
>


--
Regards,
Hiram

Mime
View raw message