problem joram connector for jonas ??

Hi every body, 

I tried to deploy a centralized JMS architecture of joram (embedded in Jonas).
And i found that there're some incoherent things ...

My architecture is composed by 3 Jonas (on a physical machine). One of Jonas
embedded a joram persistent (S1), and others Jonas inject messages into queues
of that joram:

jonas	 jonas	     jonas
S0 ------> S1 -------> S2
	   joram

The joram (embedded by jonas S1) have 3 queues and each queue have nbMaxMsg=10;
queue1 (used by MDP of jonas S0), queue2 (used by MDP of jonas S1), queue3
(used by jonas S2). The workflow goes from left to right; data flow comes in
jonas S0 and go out through jonas S2.

Configurations and tests seem to be OK. But when i wanna to take some charge
testing, i found that there's a lot of lost messages. And when i check out the
jonasAdmin console, practically 50% of messages has been moved in to a "dead
queue".

Nonetheless, i did not have any configurations for any dead queue ! 

So i would like to understand how it can occur, and where're messages that
leaved into that dead queue ? 

Thanks for your help.

st