On Thu, 2007-01-18 at 17:03 -0500, Simon P. Ditner wrote:
> It all sounds suspiciously like NAT/Firewall traversal issues. Perhaps
> double-check that canreinvite=no is set in your sip.conf for your phones?

Nope, no firewall or NAT traversal here. canreinvite was already set to
no.

> This likely falls into the "extreme" category of troubleshooting
> techniques, but my advice is to go grab this application, and leave it
> running on your asterisk servers (with an appropriate script to purge old
> data), and gateways if possible:
> 
>   http://sourceforge.net/projects/psipdump/
> 
> It's a small program that captures the SIP and RTP data for calls and
> saves it into unique caller-callee-timestamp-sessionid files in pcap
> format.

Great tool!! I added an rpmspec to the project and submitted it. I don't
see why one wouldn't install this by default on every system.

As a reminder: It's the Polycom users that are not hearing callers.

I've got a trace from good and bad calls and I can't see a difference.

I do have some questions about what I'm looking at:

The RTP Sequence does not start at 0, it's numbers as large as 50,000 or
more, it seems random. Is this DOS prevention?

The RTP Timestamp on packets sent from Asterisk start at 160 (ms I
presume), but the Polycoms seem to have it start at large numbers like
4,220,412,281, the working call had a 2,072,576,867 starting sequence.
Could this be a signed integer problem?

There is a lag between when Asterisk start with RTP and the Polycom.
There a 1280 ms delay before Polycom starts sending back RTP. The failed
call has a 1440 ms delay.

The RTP Timestamp on all packets shows a 160 ms increment between
packets, but the network trace shows about 200ms between packets.

And some other facts:

The media stream on both sides show ITU-T G.711 PCMU so I  don't think
this is a codec issue.

Help!!
-- 
John Van Ostrand                       Net Direct Inc.
CTO, co-CEO                   564 Weber St. N. Unit 12
                                  Waterloo, ON N2L 5C6
[EMAIL PROTECTED]                     ph: 518-883-1172 x5102
Linux Solutions / IBM Hardware        fx: 519-883-8533

Reply via email to