[asterisk-r2] Too many network congestions

[Moises Silva]

On Fri, Jun 5, 2009 at 11:06 PM, Daniel A.
Veiga<dveiga at advtechnology.com.ar> wrote:
>    Looking into the problem I found that I received a "network
> congestion" when the other end detected we stopped sending DNIS for
> more than about 200ms. My end never stopped sending, the problem is
> the other end is not detecting them for some reason.

How do you know the other end does not detect the tone? I read your
past e-mail: http://lists.digium.com/pipermail/asterisk-r2/2009-June/000915.html,
however, you should avoid attaching files, instead use pastebin, I
cannot see the trace because I deleted your e-mail since I was busy at
that moment. However if you had used pastebin I could access to it
now.

Just from reading your past e-mail without seeing the traces now, I
can tell you I don't think the other end is shutting down the call
with network congestion just because in 300ms they were not able to
detect the tone, sounds like a very small time frame for detection,
that kind of timeouts are typically in the order of several seconds,
not milliseconds. Anyways, please pastebin the call trace if you want
me to take a look at it.

>
>    I have no CRC errors, so I must believe the problem is not at the
> interface layer, then it must be logical (data carried by the
> interface level). I then looked at r2engine.c and found something
> strange: When transmitting DTMF then higher tone is always
> transmitted louder than the lower one. This is called twist and is
> inherited from analog line's frequency response. But the twist,
> which is implemented in openr2 library, is not used. The constant
> DTMF_NORMAL_TWIST is defined, but when openr2_dtmf_tx_set_level()
> is called it is not used. In fact a 0 is passed as the twist
> parameter.
The tones that are sent during call setup are *NOT* DTMF tones in your
case. The DTMF tones you see that are sent in openr2 library are some
experimental code for DTMF-R2 variant used in Venezuela, but Argentina
variant and all other variants use MF tones, that is, openr2_mf_tx()
etc. So this twist has nothing to do with your issue.

-- 
Moises Silva
Software Developer
Sangoma Technologies Inc. | 50 McIntosh Drive, Suite 120, Markham ON
L3R 9T3 Canada
t. 1 905 474 1990 x 128 | e. moy at sangoma.com