[SR-Users] [sr-dev] testing phase for 3.1

Timo Reimann timo.reimann at 1und1.de
Mon Aug 30 18:07:17 CEST 2010


Timo Reimann wrote:
> "Iñaki Baz Castillo" <ibc at aliax.net> schrieb:
>> 2010/8/18 Carsten Bock <lists at bock.info>:
>>> Hi, I've noticed, my RTP-Proxy changes where for the latest K
>>> already. I have modified them, to work with the latest
>>> rtpproxy/nathelper split. I will do some more testing.... (they
>>> are in the branch carstenbock/rtpproxy). I have added the
>>> suggested extension to the K-Dialog-mi-Interface: I have created
>>> a method called "mi_terminate_dlgs" which allows to terminate
>>> either all calls (e.g. before a proxy shutdown) or a specific
>>> dialog by providing call-id and optionally the fromtag (in branch
>>> carstenbock/dialog). These changes require a little more testing,
>>> which i wil do in the next days, but probably will only be able
>>> to finish by next week...
>> Hi, it sounds very good. Just take into account that current dialog
>>  module can only terminate dialogs in "stablished" status and it
>> cannot terminate early-dialogs. So it's required that your new MI
>> function checks whenever dialog state is 4 or 5. Timo fixed this
>> issue some weeks ago in dld_end_dlg MI function.
>> I will test it when I finish my holidays :)
> Being on vacation too right now, I cannot add much to this topic at
> the moment but this: If I remember correctly my fix should operate at
> a lower call hierarchy level and therefore be effective for Carsten's
> new feature automatically. However, this should be double-checked
> which I can do as soon as I switch off tourist mode :).

AFAICS, Carsten's function calls send_bye_all() which in turn calls
send_bye() which implements the "no BYE requests for non-confirmed
dialogs" protection.

Concluding, non-confirmed dialog termination should be effective out of
the box.



More information about the sr-users mailing list