[dns-operations] Accurately identifying glue records

[Roy Arends]

On May 16, 2006, at 10:33 AM, Peter Koch wrote:

> On Mon, May 15, 2006 at 11:35:15PM +0200, Roy Arends wrote:
>
>> Does it judge the 'properness' of glue or the non-existence of it?
>> (not that it needs to, just curious).
>
> first off, it depends on the parent to respond with "glue" to an  
> explicit
> A RR query. Modern parents would send a referral.
>
>> which happens to reside on nameservers with a 'org' name, glue should
>> not be included. Not even when that same server is authoritative for
>> that 'org' name. This out-of-bailiwick glue is poisonous and correct
>> resolvers must ignore it. It only leads to inflated packets.
>
> Well, that's about additional information, not "glue".
>
>> Does it really matter where the glue in the additional section comes
>> from ? Even when its through cached material instead of glue included
>> in the zone. As long as the glue is in-bailiwick, the resolver can't
>> tell.
>
> The question should be phrased differently: "Does it really matter  
> where the
> data in the additional section originates from?" (because if you  
> call it glue,
> you've already answered the question.

I guess I wasn't wearing my DNS-DOGMA hat today.

Roy