Hi Nicolas
Your patch doesn't work (no rport is added in the P-CSCF's own Via),
but apparently it isn't the patch I wanted.
The problem is that my client doesn't put rport in the Via header it
sends, so LittleIMS's P-CSCF doesn't fill in the port with "rport=..."
in that Via. On OpenSIPS, this can be worked around using the function
force_rport() (see
http://www.opensips.org/Resources/DocsCoreFcn#toc116)
which forces "rport=..." to be added to the Via even if the client
didn't ask for it. I've written a patch that does this on Cipango and
will be submitting it soon.
Regards
Damjan
> Hello Damjan,
> Thanks for the patches, I have applied them.
>
> Effectively, cipango does not support the RFC 3581 when sending requests. I have
> prepared a patch but I have not the time for testing it, could you test it ?
>
> Regards,
> Nicolas
>
> Le 06/06/2012 10:49, Damjan Jovanovic a �crit :
>
>
>
>
>
>
>
> > Hi
>
> > I've been playing around with LittleIMS, and have recently reported
> > several problem that I've found and fixed:
> >
http://jira.cipango.org/browse/LITTLEIMS-15
> >
http://jira.cipango.org/browse/LITTLEIMS-16
>
> > I've also noticed that the P-CSCF appears not to support RFC 3581,
> > that is, it doesn't add "rport=..." to Via headers sent to the I-CSCF,
> > meaning that later it can't correctly route responses back to clients
> > that are behind a NAT.
>
> > Are you guys aware of this and do you have any plans to add it? Where
> > would I start if I want to write a patch? Would this feature be added
> > to Cipango or LittleIMS?
>
> > Thank you
> > Damjan
>
>
>
> rport.patch
> 1KViewDownload