[asterisk-users] Negotiates g729 but RTP contains g711

David Herselman dhe at syrex.co
Wed Sep 23 09:17:03 CDT 2020


We have a scenario where inbound calls from an upstream provider (chan_sip) sent downstream (chan_iax2) negotiates only g729 yet RTP media contains g711. Both the upstream and downstream trunks are limited to only offering g729 whilst the initial invite from our upstream provider offers both g711 and g729. Asterisk presumably simply forwards the media from iax2 trunk encapsulation to sip encapsulation. Most calls surprisingly work, presumably by the caller's system identifying the incoming media as g711, whilst very few callers don't hear the IVR prompt. The downstream is unfortunately not within our control but can't be anything other than Asterisk, considering it's using iax2 in trunk mode.

We are running Asterisk 16.13.0, not sure what version the downstream is using.

caller -> upstream -> us -> downstream (IVR)

Herewith the SIP portion of the call, between upstream and us:
Available here: https://ibb.co/jRGvvVc

Wireshark unfortunately still cannot dissect iax2 trunk captures though, so I didn't know how to conclusively identify where this problem originates. I do however have a concern that the media we are receiving's packet size (74 bytes) indicates that it is most likely G729.

Herewith the IAX2 trunk portion of the call, between us and downstream:
Available here: https://ibb.co/r07PkkK

ie: We appear to have a reproducible environment where an inbound SIP trunk call sent to a downstream IAX2 trunk negotiates g729 in all 4 streams, receives g729 media from downstream iax2 trunk but then transmits g711a upstream.

I'm however struggling with the downstream pcap, to establish what's different about these calls. Trunk config and forwarding structure works the identical way for 50+ other flows on the same host.

David Herselman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-users/attachments/20200923/f1dfd958/attachment.html>

More information about the asterisk-users mailing list