in vs. out

Alan DeKok aland at
Fri Oct 5 09:18:42 CEST 2007

wlan at wrote:
>>   You don't have to interoperate with broken vendors.  You tell users to
>> throw the equipment away, and to buy working equipment.
> For some, that is not very economical - nor environmentally friendly :)

  You can buy equipment that works, or equipment that doesn't work.
It's a choice.

> I started the list; sorry, I couldn't help myself and started it here:

  <sigh>  You started asking questions on the FreeRADIUS list.  Rather
than updating the FreeRADIUS Wiki, you've gone somewhere else.  Nice.

> I took a perhaps less hostile classification than 'broken' - and labeled
> them as having a "AC" or "Client" perspective.

  Perhaps "non RFC compliant" would be a politer choice of words.

> Of course, the Access
> Controller/NAS  is the right meaning as defined in this forum.

  Perhaps you haven't been reading my messages.  This forum is NOT
defining anything.  The definitions existed for years before FreeRADIUS
existed.  Stop trying to claim it's a disagreement between vendors.
It's not.

?> As you
> can see, there is a bit of an issue. In a Gemtek manual, they mention
> the problem - explaining the Client and AC point's of view, but still
> ultimately defaults to the Client perspective with an option to reverse.
> As for coova-chilli, I actually (yes, I'm ready for the public
> ass-kicking) changed the accounting to be like that of Gemtek - with the
> option to toggle. At the time, my objective was purely compatibility
> with back-ends already built for some of the vendors in this list.

  The default should be the correct meaning.  The documentation should
have big bold warnings that vendors doing it the other way are

> Huh, I wonder if there was something originally 'lost in translation'

  The original RFC's were written the the designers and implementors of
RADIUS, in Ann Arbour, Michigan, USA.  They might be poorly phrased, but
there is no confusion.

> with how this got implemented. With some Googling, I came across:
> Which states for Acct-Input-Octets: "Volume of the downstream traffic of
> the User" and Output-Octets with "upstream traffic of the user". That
> sounds rather Client centric -- it's not to / from the User, for
> instance. Are we expecting too much from the (off-shore) out-sourcing
> companies? :)

  Apparently people implementing NAS equipment can't be bothered to read
the spec, or to ask questions if they don't understand it, or to see
what the large vendors do.

  Alan DeKok.

More information about the Freeradius-Users mailing list