inline

On Mon, 2003-09-08 at 08:34, Adi Even-Tzur wrote:
> thanks for the response. 
> 
> I read about the second way to deal with asymmetric record routing by
> setting multiple record-route headers by the proxy.
> my question is whether this is going to be a substitute solution that is
> recommended by rfc 3261.

Yes. 

> 
> In addition, I'm still confused about the current solution described in rfc
> 3261 section 16.6.4 and 16.7.8.
> The example you gave is when a request is received over TLS and sent over
> UDP.
> 
> You've written that changing the record-route from SIP to SIPS when the
> response is received is not the correct behavior 
> and the proxy just need to keep the record-route addresses to be distinctive
> enough so that they will resolve to the correct interface.  
> 
> but, according to rfc 3261:
> 1. when handling the request according to section 16.6.4, rfc 3261:
> "a proxy that receives a request over TLS, but generates a request without a
> SIPS URI in the Request-URI or topmost Route
> header field value (after the post processing of bullet 6), MUST insert a
> Record-Route header field that is not a SIPS URI."
>  - meaning the proxy must insert a record-route with a URI that is not a
> SIPS.
All of that come after and is part of the clause
"If the Request-URI contains a SIPS URI, or the topmost Route header
 field value (after the post processing of bullet 6) contains a
 SIPS URI,"

I've added a bug (734) against the spec to make it more clear that the
rest of that paragraph is dependent on that starting sentence.

> 2. when handling the response according to section 16.7.8, rfc 3261:
>  "If the proxy received the request over TLS, and sent it out over a non-TLS
> connection, the proxy
> MUST rewrite the URI in the Record-Route header field to be a SIPS URI." 
> - meaning the proxy need to rewrite the record-route with URI that is not a
> SIPS to one with SIPS. 

This is a bug in the spec. The conditional here should be the same as
what I quoted above (noting that its what was in the request associated
with this response that's being tested). This is being tracked now as
bug 735. Thanks for pointing it out!

> 
> Is the rfc not accurate, or am I missing here something ?
> 
> regards,
> 
> Adi Even-Tzur.
> 
> -----Original Message-----
> From: Robert Sparks [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, September 02, 2003 5:45 PM
> To: Adi Even-Tzur
> Cc: [EMAIL PROTECTED]
> Subject: Re: SipIt Interop problem - when proxy should rewrite URI of
> Record-R oute header
> 
> 
> First off -
>   Using TLS and using a sips: URI are not the same thing.
>   sips: implies TLS.
>   TLS does not imply sips:
> 
> In the scenario you describe below, there is no reason for a
> sips: URI to spring into existence. In fact, it would be an
> error if one did. A sips: URI says something about the security
> of transporting messages _all the way to the far endpoint's domain_,
> not just the next hop.
> 
> The real concern in this case is providing enough information in the
> RR header field so that subsequent requests from each direction show
> up at the interface the proxy wants them to show up. Assume that what
> the proxy wants is to continue using TLS upstream and the non-TLS (say
> UDP) downstream. It needs to give its downstream proxy a URI that 
> resolves to UDP according to 3263. It needs to give its upstream proxy
> a URI that resolves to TLS. 
> 
> RjS
> 
> btw - A better way to deal with this asymetric record routing than
> rewriting the fields in responses was identified several months ago
> on the SIP list. Check out 
> 
> http://bugs.sipit.net/sipwg/show_bug.cgi?id=664
> 
> 
> On Sun, 2003-08-31 at 05:30, Adi Even-Tzur wrote:
> > Hello,
> > 
> > This is an issue that was raised during SipIt 13 "sips spirals" advanced
> > test.
> > 
> > When a proxy received a request over TLS, but the Request-URI is without
> > "sips", and the proxy sent the outgoing request
> > over non-TLS - should the proxy rewrite the URI of the Record-Route header
> > it set when forwarding the response from "sip" to "sips" or should it
> leave
> > it as is ?
> > 
> > In other words, does the transport type the request was received and sent
> is
> > what that matters or is it the URI type that matters ?
> > 
> > 
> > regards,
> > 
> > Adi Even-Tzur.

_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to