I have an odd problem that I have not been able make any progress on.
Freepbx 2.4.1.1
Asterisk 1.4.21.2
Running on debian etch, also tried on centos 5.0. Same problem.
Every 4th or 5th call into a queue, the caller cannot hear the agent. The Agent can hear the caller. With recording enabled, both sides are recorded, the caller grumbling because they just hear silence (or actually a very quiet buzz, occasionally switching to a loud buzz after a minute or too in my testing. Callers never stay on that long.) and the agent saying “Hello. Hello. Hello?”
Inbound calls are coming from our provider as SIP over local ethernet.
I’ve done tcpdumps and reconstructed the RTP stream and on my internal LAN I hear both voices. On the external interface I do not hear our agent, just the caller. So it seems the stream is dropping in the server/asterisk/freepbx somewhere. This only happens on calls to the queue and pretty regularly. The path is: Time Condition-> IVR -> Queue -> agent. moh and call recording are enbled, but I have tried disabling those as well.
Does anyone have any specific ideas about what could cause this or where I should focus debug efforts to isolate the problem?
With the excellent (paid) support help of the freepbx team, this problem has been narrowed somewhat but not solved. It does not appear to be caused by or related to freepbx or *
All RTP is flowing the way it should and a capture of the packets that we are sending towards our provider shows that we are sending them the proper stream of data. SIP and codecs look correct. So it looks now like a random problem on the providers end. They are working to isolate it.
An even odder issue is that after about 3 minutes on the call where the caller cannot hear the agent, the caller hears a very loud buzz which eventually distorts and breaks down into a number of harmonics. I’ll post updates as I know more.
No change in this. The provider is stumped. We’ve changed PRI’s and upstream switches. We are directly connected to a lucent box for RTP and OpenSER for sip.
It seems evident that something between * and the lucent box is getting confused. All sip transactions on working and non-working calls are the same.
You might want to get out your trusty (or rusty) oscilloscope and take a look at noise on your ethernet cabling. I have had phones get a buzz on them in one installation. I beat my head on the wall for months trying to figure out why, no matter which phone I put at this one station, it would fail with humm on it.
Then I took out the oscilloscope and looked at the noise on the ethernet cable. I had right at 0.75 volt, peak-to-peak noise. The funny think was that the waveform was triangularly shaped. As soon as I turned of the adjacent UPS, the noise went away.
This VOIP stuff will tolerate a fair amount of noise on ethernet cabling, but once that threshold is reached, voip phones do unexpected things.
Thanks for the idea. I might be forced to check that out next. The behavior is odd though. Its not tied to any location or station cable.
It seems to occur more frequently on inbound calls and on some outbound calls. This is often reported as “DTMF does not work” but its really that the RTP is getting dropped by the peer we are talking to. wireshark can always put the stream back together.
The provider has loaded new code on a Lucent box and we plan to test this week. It seems the lucent might be having trouble with some ulaw codec rtp streams. We’ll see.
My inbound teliax line has been working well to provide inbound calls to a support queue.
The problem has been resolved by an upgrade of code on the Lucent box the provider uses. I can get details if anyone would like to know the previous and current versions.
it would be nice to have that posted here so that other who go looking months from now will know. If there is a specific bug reference that fixed it in the firmware that would be nice to post also so that others can call there provider and say here is the formal bug notice and I need firmware that fixes it.
4 years after this post… I am having a similar issue. I am running asterisk 1.8. FPBX 2.10 My call flow:
Windstream Sip trunk -> Public IP on eth0 A2Billing server running * 1.4 private IP on eth1-> Client server with *1.5 FPBX 2.10 registered with no nat -> Inbound route -> call flow control -> time condition -> receptionist queue - receptionist -> transfer to queue with iSympnony -> calls picked up from queue have no audio or sometimes one way audio.
Client devices have no nat settings as the public/private is handled by the a2billing proxying the call.
I am not using Call Recording.
Phones are cisco 7940 auto provisioned from device manager. I have changed the re invite settings off and on in the extensions screen with no difference (I am not sure cisco 79x0 support re-invites on sip)