Hi Elodie, Jeromi,

I did a test a similar to yours. But without MTOM attachments and using the Userguide sample in Sandesha2. I tested  SyncPuing and AsyncEcho scenarios with network failures (by unpluging the network cable in the middle of the transaction etc.).

There was a bug in the Sender thread pool logic of which prevented Sandesha2 from sending some messages at times. after fixing this things ran fine.
Sandesha2 has a TestCase that tests message retransmissions (MessageRetransmissionTest). But this error was due to a race-condition and was not covered in the test.

Can you try again with a Sandesha2 update. If the problem still occurs please try without the MTOM attachment.

Chamikara


On 9/7/06, Jerome Camilleri <[EMAIL PROTECTED]> wrote:
Hi,

Our purpose with Elodie is to test WS-RM with axis2.
Elodie successfully used sandesha in a local network when there isn't
messages lost.
Now, we would like to test the RM when the network lose messages.

After sending and receive acknowledgment for sequence number 3 the
network cable was unplung (to simulate a network problem).
After one minute the cable is replugged.

At this moment, the sequence number 7 is send.
The specification said that
"The RM Source MUST assign eatch reliable message a sequence numer
beginning at 1 and increasing by exaclty 1 for eatch subsequent reliable
message"

Sequence 8 was send and we received acknowledment response with to
AcknowledgmentRange(1,3) and (7,8)

It seems that the terminate sequence of the first sequence was never send ?

A just have a look at your SVN source code and I found scenarios tests
for MTOM, security and so one but nothing about retransmition...

We using Axis1.0 (May 2006) and a snapshot version of sandesha (middle
of Aout).

Regards,
Jérôme


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


Reply via email to