[Devel] RADIUS Dictionary

[Peter Nixon]

On Tue 14 Nov 2006 02:33, Peter Nixon wrote:
> On Tue 14 Nov 2006 01:02, you wrote:
> > Peter Nixon wrote:
> > > Why is it that neither SER nor OpenSer has submitted the SER radius
> > > dictionary for inclusion in FreeRADIUS??? I notice all the
> > > documentation explains how to copy it in, but the correct solution
> > > would be to have it as a vendor dictionary and be done with it. Does
> > > anyone have anything against me committing it to FreeRADIUS cvs? I also
> > > notice that it doesn't currently have a vendor number from IANA at the
> > > top (has everyone just been too lazy to apply for one?). The author of
> > > the radius module should apply for one, or alternatively we can just
> > > use mine (26514)
> > >
> > > We have a new stable FreeRADIUS release due out in the near future
> > > (days or weeks at most) so a quick decision will allow for inclusion in
> > > what should be the final release of the 1.1x branch.
> > >
> > > Waiting for replies/flames before I do anything...
> >
> > I think this is a great idea that will reduce a lot of installation
> > hassles.
>
> Can anyone tell me why the Sip-Method Values in the dictionary with
> OpenSER:
>
> ### Sip-Method Values ###
> VALUE Sip-Method         Undefined      0
> VALUE Sip-Method         Invite         1
> VALUE Sip-Method         Cancel         2
> VALUE Sip-Method         Ack            4
> VALUE Sip-Method         Bye            8
> VALUE Sip-Method         Info           16
> VALUE Sip-Method         Options        32
> VALUE Sip-Method         Update         64
> VALUE Sip-Method         Register       128
> VALUE Sip-Method         Message        256
> VALUE Sip-Method         Subscribe      512
> VALUE Sip-Method         Notify         1024
> VALUE Sip-Method         Prack          2048
> VALUE Sip-Method         Refer          4096
> VALUE Sip-Method         Other          8192
>
> Dont match the values in
> draft-schulzrinne-sipping-radius-accounting-00.txt:
>
>              0  INVITE
>              1  BYE
>              2  REGISTER
>              3  CANCEL
>              4  OPTIONS
>              5  ACK
>              6  SUBSCRIBE
>              7  NOTIFY
>
> Not only does OpenSER have more values, but the integers dont match the
> same values in the spec....

Oh god. it gets worse. 

http://www.openser.org/docs/openser-radius-1.0.x.html documents a "string" 
attribute "Sip-Src-IP" instead of using schulzrinne's "Sip-Source-IP-Address" 
which is an "ipaddr" octet field. Does anyone know why openSER follows the  
schulzrinne draft in places, implements competing attributes in others and 
redefines existing values in others???

Additionally dictionary.radius which is included in openser-1.1.0 does not 
match with the documentation at 
http://www.openser.org/docs/openser-radius-1.0.x.html

I have added dictionary.schulzrinne-sipping to FreeRADIUS cvs. It follows 
draft-schulzrinne-sipping-radius-accounting-00.txt

All of the relevant Digest attributes (previously draft-sterman-aaa-sip, now 
RFC 4590 are already in FreeRADIUS)

See http://wiki.freeradius.org/RFC for more detail.

Anything that doesn't match the draft(s) or RFCs in openSER needs to be sent 
as a VSA (http://wiki.freeradius.org/VSA). If the maintainer of the RADIUS 
code in openSER could contact me (either on or off list) I would be happy to 
explain how this is done.

Cheers
-- 

Peter Nixon
http://www.peternixon.net/
PGP Key: http://www.peternixon.net/public.asc