Best Practice for Real-Time Media Forking/SIPREC in an Inline B2BUA Hairpin Setup (KEMI/Python)

--===============0268952071==
Content-Type: multipart/alternative; boundary="0000000000003c13d8065588380f"

--0000000000003c13d8065588380f
Content-Type: text/plain; charset="UTF-8"

Hello Kamailio Community,

I am looking for the best practice or native method to implement real-time
media forking/forwarding directly from our proxy/media layer to an external
Python/FastAPI service, keeping the architecture completely clear of any
intermediary PBX engines.
Our Environment

   -

   *OS:* Ubuntu 24.04 LTS
   -

   *Kamailio Version:* 6.1.3 (x86_64/Linux) running the Python KEMI engine (
   kamailio.py)
   -

   *RTPEngine Version:* 26.2.0.0+0~mr26.2.0.0

Our Scenario

We have a fully working signaling hairpin configuration handled inside
kamailio.py. A call arrives from our carrier gateway, loops through our
proxy framework, and terminates to the customer gateway, with media
successfully anchored via RTPEngine.
The Objective

We need to passively replicate/fork a twin stream of the live full-duplex
voice packets directly over the network via UDP or TCP to an asynchronous
socket listener on our backend Python application. We want this to happen
completely in real-time, with zero disk-spooling/file-writing dependencies.
Our Questions:

   1.

   *RTPEngine Network Replication:* Given that we are on RTPEngine 26.2, is
   there a native control flag or parameter inside the daemon protocol to
   replicate and push live UDP streams straight to an external network target (
   IP:Port)?
   2.

   *Kamailio Native Alternatives:* If RTPEngine cannot forward packets
   directly over the network to a raw socket, are there specific Kamailio
   6.1.3 modules or KEMI-exported functions we should utilize to capture and
   mirror the live RTP streams down to our Python service?

We are trying to keep this pipeline as lightweight and bare-metal as
possible (Kamailio + RTPEngine + Python). Any guidance, architectural
patterns, or module recommendations to achieve this clean
proxy-to-application streaming would be immensely appreciated.

Thank you for your time!

--0000000000003c13d8065588380f
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><p>Hello Kamailio Community,</p><p>I am looking for the be=
st practice or native method to implement real-time media forking/forwardin=
g directly from our proxy/media layer to an external Python/FastAPI service=
, keeping the architecture completely clear of any intermediary PBX engines=
.</p><h4>Our Environment</h4><ul><li><p><b>OS:</b> Ubuntu 24.04 LTS</p></li=
><li><p><b>Kamailio Version:</b> 6.1.3 (x86_64/Linux) running the Python KE=
MI engine (<code>kamailio.py</code>)</p></li><li><p><b>RTPEngine Version:</=
b> 26.2.0.0+0~mr26.2.0.0</p></li></ul><h4>Our Scenario</h4><p>We have a ful=
ly working signaling hairpin configuration handled inside <code>kamailio.py=
</code>. A call arrives from our carrier gateway, loops through our proxy f=
ramework, and terminates to the customer gateway, with media successfully a=
nchored via RTPEngine.</p><h4>The Objective</h4><p>We need to passively rep=
licate/fork a twin stream of the live full-duplex voice packets directly ov=
er the network via UDP or TCP to an asynchronous socket listener on our bac=
kend Python application. We want this to happen completely in real-time, wi=
th zero disk-spooling/file-writing dependencies.</p><h4>Our Questions:</h4>=
<ol start=3D"1"><li><p><b>RTPEngine Network Replication:</b> Given that we =
are on RTPEngine 26.2, is there a native control flag or parameter inside t=
he daemon protocol to replicate and push live UDP streams straight to an ex=
ternal network target (<code>IP:Port</code>)?</p></li><li><p><b>Kamailio Na=
tive Alternatives:</b> If RTPEngine cannot forward packets directly over th=
e network to a raw socket, are there specific Kamailio 6.1.3 modules or KEM=
I-exported functions we should utilize to capture and mirror the live RTP s=
treams down to our Python service?</p></li></ol><p>We are trying to keep th=
is pipeline as lightweight and bare-metal as possible (Kamailio + RTPEngine=
 + Python). Any guidance, architectural patterns, or module recommendations=
 to achieve this clean proxy-to-application streaming would be immensely ap=
preciated.</p><p>Thank you for your time!</p></div>

--0000000000003c13d8065588380f--

--===============0268952071==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

__________________________________________________________
Kamailio - Users Mailing List - Non Commercial Discussions -- [email protected]
To unsubscribe send an email to [email protected]
Important: keep the mailing list in the recipients, do not reply only to the sender!

--===============0268952071==--