SIP Provider calls to CUE disconnect at 30 seconds into message recording

Kris Thompson · ‎05-04-2009

Calls inbound from SIP Provider to CME/CUE (B2BUA) are disconnect at 30 seconds into voicemail.

My understanding is that RTCP reports should be sent from the RTP endpoints to keep the RTP session alive in this kind of situation (one direction is muted).

At 30 seconds into leaving a voicemail, we receive a "BYE" from the SIP provider.

Running debug voip rtcp, and even ip packet shows NO RTCP packets being sent from the Router - 12.4(24)T

Looking at "show call active voice brief" shows that there is no increase in the RX traffic on the CUE call leg while the voicemail message is being recorded.

17B0 : 495 590510970ms.1 +90 pid:2 Originate 300 active

dur 00:00:36 tx:1805/288800 rx:681/108483

IP 172.16.2.2:16906 SRTP: off rtt:0ms pl:0/0ms lost:0/0/0 delay:0/0/0ms g711ulaw TextRelay: off

media inactive detected:n media contrl rcvd:n/a timestamp:n/a

long duration call detected:n long duration call duration:n/a timestamp:n/a

</code></pre>

CiscoIOS was recently upgraded to 12.4(24)T. CUE is v2.3(2).

This router is also performing NAT between the SIP provider and the Voice subnet.

1) Is it up to CUE to initiate all RTP and RTCP traffic with the B2B-UA just passing along the packets? What is the expected or required RTCP/RTP behavior in a B2B-UA environment?

2) How can I confirm if CUE is sending RTCP packets and if they are received and forwarded by CiscoIOS.

Any help would be much appreciated.

Kris Thompson · ‎06-02-2009

Still no progress has been made on this problem. Today I created a new dial-peer without B2BUA and performed a packet trace. No RTCP traffic is seen generated by CUE. Isn't RTCP a requirement of the RFC for RTP? The SCCP leg stays alive due to SCCP traffic, but what about the leg from the CUE to a SIP provider. Surely someone has seen this issue before.

Does CUE send RTCP packets? Can they be enabled?

Kris Thompson · ‎10-05-2009

bueller?

This problem still exists. There seem to be some suggestions to use "voice-class source interface loopback x" on the dial peer, but we are not using Loopback addresses.

As posted here:

http://uc500.com/connecting-uc520-sip-trunk-itsp-internet-telephony-service-provider-nexvortex#comment-2485

I also found this:

https://www.myciscocommunity.com/docs/DOC-1445#Cannot_leave_voicemail_for_more_than_30_seconds_

Which also seems to address this, but the instructions pointed to are pretty awful and cannot be followed.

Can someone please explain what is required to make RTCP packets flow from CUE to the SIP provider?

Nicholas Matthews · ‎10-06-2009

The only reason you would need the voice-class source command is if your NAT wasn't working correctly and you needed to source the packets the router was generating from a loopback interface.

We commonly see these as causes of the 30 second audio problems:

IOS FW. SIP Inspect tears down the call. You can increase the IOS FW timer for SIP or UDP and this may help.

Refresh from SIP provider. Quite often the refresh causes the flow to become more complex, and a bug or failure to negotiate tears the call down.

RTCP media inactivity timers. I don't think CUE supports RTCP.

You can debug ccsip messages and find out which side is disconnecting, which would be helpful to narrow down the troubleshooting.

-nick

oliverpowell · ‎05-22-2012

Hi Kris,

Did you ever resolve this issue?

We are having the same problems here with our customer.

30 seconds into leaving a message we recieve a BYE from the SIP providor and that call disconnects.

Any help appreciated.

Oli

Kris Thompson · ‎05-22-2012

To my knowledge, this issue was never resolved. The SIP Provider wanted to do RTCP and the CUE (or IOS in B2B-UA) apparently does not support it.