Re: [dhcwg] [radext] draft-ietf-dhc-dhcpv6-radius-opt-10

[Jouni Korhonen <[e-mail]>]

On Apr 5, 2013, at 2:42 PM, Ted Lemon <[e-mail]> wrote:

> On Apr 5, 2013, at 3:19 AM, Jouni Korhonen <[e-mail]> wrote:
>>  The option-data of OPTION_RADIUS is a list of one or more RADIUS
>>  attributes received in the Access-Accept message from the RADIUS
>>  server. The OPTION_RADIUS can only contain RADIUS attributes
>>  listed in the IANA Registry of 'RADIUS attributes permitted in
>>  DHCPv6 RADIUS option'.
> 
> So you took out the normative language, right?   Was that intentional?

That was intentional. If that is a concern one can always change the
"can only" to "MUST". That works too, since the previous sentence
already points out that the option contains one or more attributes,
not all.

>> The next question I have is what happens when a relay includes an attribute
>> that the server does not understand or is not listed in the registry? There
>> is no versioning thus it is possible that relay and server have a different
>> understanding what the IANA registry is. Now the text in Section 6 only
>> addresses the case where the server does not understand the DHCP option.
> 
> Good question.   I think the right answer is that that RADIUS attribute is silently ignored, because, as you say, the server might not be up to date.

Blindly dropping an attribute might not work in all cases. For example, in
some cases the server might not then be able to provide all information
the relay needs.. That is more of a DHCP specification issue but what I
would like to see in this I-D is some text pointing out that the server
and the relay may have a different idea of the registry and the protocol
design need to take that into account.

- Jouni




>