Thank you Richard.

Maybe Alex can tell us what amount of stress were put on rtpengine to make it crash. 

Regards,
Carlos


On Tue, Jul 1, 2014 at 1:57 PM, Richard Fuchs <rfuchs@sipwise.com> wrote:
On 01/07/14 12:59 PM, Carlos Ruiz Díaz wrote:
Hi Richard, Alex,

I'm curious about it. Is there a way of simulating this? Under what
conditions should this happen?

The first case described (differing number of ports) I'm not 100% sure about. I'm thinking the little used slash notation in the SDP m= line might be to blame... one side using it (say, 1234/2 to advertise 4 ports being used), the other side not using it. You'd end up with 4 ports open for each side, but some of them unused. And then UDP packets coming in on those ports.

The race condition is harder to test. You'd have to predict which ports will be opened when an "offer" is received, and have these ports hammered with UDP packets ahead of time, while the "offer" is being processed.


cheers

_______________________________________________
sr-dev mailing list
sr-dev@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev



--
Carlos
http://caruizdiaz.com
http://ngvoice.com
+52 55 3048 3303