[asterisk-users] Asterisk 1.8.7.2 now sends rport always

[Kevin P. Fleming]

On 12/18/2011 01:42 AM, José Pablo Méndez Soto wrote:

> I have been testing with Cisco phones and have been able to register
> them with new firmware 9.2.1 (7911/7945/7970). All worked until I
> realized that from version 1.8.7.2, the VIA header contains the rport
> parameter, which breaks the phone registration process. Basically, the
> device can´t parse the VIA header this way, and when it gets the 200 OK
> to the REGISTER message containing the rport parameter, it refuses to
> process the registration internally, although it doesn´t complaint about
> it and Asterisk shows it as registered.

First, let me say that it is pretty ridiculous that Cisco phones refuse 
to accept SIP responses with rport parameters in the Via header. But 
getting back to your problem... did you read the CHANGES file included 
with Asterisk 1.8.7.2? The *only* change between 1.8.7.1 and 1.8.7.2 was 
specifically handling of the 'nat' option in chan_sip to address a 
security vulnerability, but your message reads as if you are not aware 
of this.

> Asterisk 1.8.7.1  doesn´t behave this way and all works fine. The
> documentation about the use of the nat= parameter in sip.conf states:
>
> ;        nat = no                ; Default. Use rport*if* the remote
> side says to use it.

This is a bug in the sample configuration file; 'no' is no longer the 
default, 'force_rport' is.

> I understand that the other side must send an empty rport parameter to
> report the far end it needs the rport field to be filled in as per the
> RFC. The IP Phone is not sending the field at all, generating
> incongruity between the documentation and the real behavior. The only
> reason I think Asterisk would find the condition to be true, is due to a
> mismatch between the source port and VIA header ip:port inside the
> REGISTER message.
>
> Could this be the trigger of the 200 OK with rport (and, other SIP
> messages as well)?

It is exactly that; 'force_rport' is now the default, and if you need 
'no' behavior, you have to explicitly configure it that way.

> Can it be implemented a nat = never option in future releases?

There is no need for such an option (which is why it was removed in the 
Asterisk 1.6.x timeframe).

-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
Jabber: kfleming at digium.com | SIP: kpfleming at digium.com | Skype: kpfleming
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
Check us out at www.digium.com & www.asterisk.org