[Spce-user] ngcpct='sip:127.0.0.1:5080' in Contact header

[John Murray]

Yes one problem with some vendors is the loopback address of 127.0.0.1 which
as you say can be fixed by changing the internal address as described in
handbook section 5.5.1.
The other problem some vendors have is with the 'ngcpct=' section of the
Contact: header as:

Contact: <sip:ngcp-lb at 10.23.11.18:5060;ngcpct='sip:10.23.11.18:5080'>'

Some vendors 'forget' this and reply to sip:ngcp-lb at 10.23.11.18:5060 and
some get stuck and don't reply at all when they see this.

Regards

John

-----Original Message-----
From: Skyler [mailto:skchopperguy at gmail.com] 
Sent: 27 March 2013 15:55
To: Andreas Granig
Cc: John Murray; spce-user at lists.sipwise.com
Subject: Re: [Spce-user] ngcpct='sip:127.0.0.1:5080' in Contact header

Hi,

On 3/27/2013 8:41 AM, Andreas Granig wrote:
> Hi,
>
> On 03/27/2013 04:11 PM, John Murray wrote:
>> With a peer you are typically in some sort of relationship and can 
>> negotiate any fixes (not always the case of course). But with a 
>> subscriber you are dependent on the code in the device.
>> In my experience with PBXs ( naming no names (but NEC is terrible)) 
>> are very bad with this header so it would be very helpful if there 
>> were some switch against the subscriber to remove this as a 
>> workaround.
>> Getting the vendor to do something about this has always proven very 
>> difficult/painful/slow.
>
> Well, the reason this has been added was to mask the local sems 
> contact of 127.0.0.1:5080, because some peers are doing far-end nat 
> traversal (like the SPCE itself), and if some peer tries to fix this, 
> the SIP flow is again broken.
>
> So in one way or another, this information needs to be coded into the 
> message. I'm open for any better suggestions of course :)
>
> Andreas
>

  I created an lo:0 with a private IP address (10.10.xxx.xxx) for lb >
b2buas: & proxies > sip:. That may help in this case as its likely the
127.0.0.1 which is confusing the upstream proxy.

Skyler