Asterisk is running on a openwrt based fiber gateway and offers voice on it for two POTS.
Asterisk has version 1.8.10.
Now my problem, the voice platform sends during a call every ten minutes a reinvite just to be sure the call is still going. I see the asterisk client replies to it every time it receives the INVITE.
All wireshark traces have RTP in both directions even if I have oneway on my phone.
And it is allways the same side that can’t hear the other side anymore, the one connected tho the asterisk client.
So a call can happen like this:
first 10 minutes all ok.
after 10 minutes, first REINVITE, sudden oneway audio
another 10 minutes, second REINVITE, self-healing both sides can hear each other
But, it is not allways the same. It can happen after 10, 20 ,30,40 … minutes. Not predictable.
I am new to asterisk and not sure if I have possibilities like asterisk on a real server.
But still, anybody out there with an idea where I have to dig?
What you think, is it an asterisk problem at all?
canreinvite is an obsolete name for directmedia. It is so often used in cook book configurations, especially those suggested by ITSPs, that I doubt it will ever be removed as an alias.
The option that was never supported was directrtpsetup.
The re-invites, in this case, are not controlled by any of these options.
It seems likely that the peer or router is broken, possibly mishandling an INVITE with no SDP.
Hello ambiorixg12
The INVITE comes from the voice platform, Cirpack, and has to be answered by the Asterisk Client.
Would in such a case directmedia=no help?
I mean Asterisk is not the one that sends INVITE, is has to reply to it.
re-INVITE is used to modify the dialog and the parameters of the session, as soon as the re-INVITE IS SENT your communication is broken, as it was said before peer is possibly mishandling it , and as Asterisk by default tries to redirect the RTP media stream to go directly from the caller to the callee, with directmedia we are just trying to turn off this
I know the trace is still outstanding.
Now I know, debugging is still possible on this device so I used it as good as I can.
So far I would say it has to be problem with the jitterbuffer.
There was a buffer destroy and rebuild message after x*10min.
In default it is set to ‘adaptive’. I switched now to ‘fixed’ and the problem seems to be gone. With that knowledge I’m not sure the trace would help in any way.
Many thanks so far.