[OpenSIPS-Users] RTPEngine Setup Delay Between External Users

[Nate Baker]

Hello Everyone,

I'm trying to solve a problem where there is a 3-4 second delay before
audio is connected when using RTPEngine between two external users.  In
this scenario OpenSIPS/RTPEngine is used as sort of an SBC for remote
users, and there is a PBX behind OpenSIPS that just proxies the call (it's
not creating a 2nd call and bridging them).  The server is multihomed (has
a public and private IP).  Trying to visualize it (excuse my poor ASCII
art):

         ___________       _________
UAC --> |           | --> |         |
        | OpenSIPS  |     |  PBX    |
        | RTPEngine |     | (Proxy) |
UAC <-- |___________| <-- |_________|

It seems like since the call ID stays the same RTPEngine is having trouble
establishing the interfaces, even when using any of the "to-tag" or
"via-branch=(1|2|extra)" flags.  So the call connects, and I get audio
after about 3-4 seconds.  During that time the RTPEngine log shows a bunch
of entries like:

INFO: [...]: Switching local interface to x.x.x.x:31000
INFO: [...]: Switching local interface to x.x.x.x:30966
WARNING: More than 30 duplicate packets detected, dropping packet to avoid
potential loop
WARNING: More than 30 duplicate packets detected, dropping packet to avoid
potential loop

After 3-4 seconds with a ton of those messages it confirms peer addresses
back and forth several times, and audio is connected fine for the rest of
the call.  The only way I have been able to get it working properly is to
use the "call-id=" flag and set a custom call-id for each.  Can anyone
recommend the way this should be handled?  Or is setting a custom call-id
the appropriate way to handle this?

Thanks,
Nate
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20200610/2e1fdfed/attachment.html>