[infinispan-dev] CommandAwareRpcDispatcher desired logic

[Dan Berindei]

On Wed, May 29, 2013 at 4:50 PM, Pedro Ruivo <pedro at infinispan.org> wrote:

> Hi all,
>
> //related to https://issues.jboss.org/browse/ISPN-3100
>
> In here [1] I've added three simple tests to the CARD logic with the
> ResponseFilter.
>
> #1 testResponses() => tests if all the responses are returned. This test
> is failing because the FutureCollactor only returns the last response.
>
> *question* is this the desired behaviour or is it a bug (and this test
> is incorrect)?
>
>
I think it was intentional, because it only really had to work with
ClusteredGetResponseValidityFilter at the time. But I'm all for changing it
to return all the (validated) responses.


#2 testTimeout() => tests the timeout when a ResponseFilter is used. It
> sets the timeout to 1000 milliseconds and sends a command that sleeps
> for 5000 millisecond before reply. This test is failing because no
> TimeoutException was thrown (as expected). That's why I've created
> ISPN-3100.
>
> *question* is the desired behaviour to wait infinitely? If yes, then
> ISPN-3100 is invalid.
>
>
I'm pretty sure that's not intentional. FutureCollator.futureDone() should
be called when a UnicastRequest finishes with a TimeoutException and it
should decrement expectedResponses. When all the UnicastRequests are done,
it should notify() the caller thread, which should throw a TimeoutException.



> #3 testTimeoutWithoutFilter() => tests the timeout without a filter.
> It's working correctly!
>
> Thanks!
>
> Cheers,
> Pedro
>
> [1]
>
> https://github.com/pruivo/infinispan/commit/1fc26d740b773bf3864298615f61ca57b964d9cf#diff-0
> _______________________________________________
> infinispan-dev mailing list
> infinispan-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/infinispan-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/infinispan-dev/attachments/20130529/01268e6b/attachment.html