Greger:
 
 The question of the day seems to be whether or not the two Cisco gateway do indeed have the abilty to speak to each other. No firewall exists. No "obvious" ACLs exist. We are looking at the control plane policing policies but they do not appear to be an issue. It has long been suspected to be a dialpeer problem but we do not see again "obvious" issue. An ngrep from the proxy perspective shows a 503 service unavailable from the gateway handling the first call leg. That seems to be the root symptom but the cause of this error remains a mystery.
 
-Steve

From: Greger Viken Teigre [greger@teigre.com]
Sent: Sunday, March 30, 2008 7:51 AM
To: Steven C. Blair
Cc: serusers@iptel.org
Subject: Re: [Serusers] Help with attended (consultative) call transfer problem

Can one Cisco gw do media directly to another gw on the SIP interface?  I.e. will you see any difference if you proxy the call through your rtpproxy?
You don't say anything about how the transfer fails.  As you have an existing media session going, it needs to be replaced and you get hairpinning of the media.
g-)

Steven C. Blair wrote:
 
Hello:
 
We have been having persistent problems with consultative (attended) call transfer and I’d like to ask the list for some input.
 
We have three Cisco 2821 gateways each with two PRIs connecting to Verizon Centrex service. The route advance sequence defined by Verizon results in all calls entering our VoIP environment entering via gateway #1. If no channels exist calls fail over to gateway #2 and finally #3.
 
For load balancing purposes our SER 0.9.7-pre3 proxy will send outbound calls to gateway #3, then #2 and finally #1.
 
We are using Cisco 79x0 phones running SIP load 7.3. All other types of calls including blind transfers work.
 
Call transfers in which both call legs traverse the same gateway work. Only transfers where each call leg traverses different gateways fail. Given the inbound and outbound routing described above it is easy to see how most transfers fail.
 
From a protocol perspective the transfer follows RFCs. I have ethereal traces if anyone needs them.
 
Cisco has worked on this problem and is now suspecting that a transfer will only work if both call legs traverse the same gateway. It is unclear if this is technically correct, a configuration issue in our gateway or a mistake.
 
To further complicate the issue calls from the PSTN, through a gateway to an IP phone which is transfers to another IP phone on the same proxy work. Failures only happen when the second call leg is to the PSTN and traverse a different gateway.
 
Has anyone on this list seen this issue or have any input whatsoever?
 
Thanks,Steve
 
 
 
 
Senior Network Engineer,
Information Systems and Computing
Networking and Telecommunications , Suite 221A /6228
University of Pennsylvania
Voice:215-573-8396
FAX:215-898-9348   
 
 
 

_______________________________________________ Serusers mailing list Serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers