> Mark.Andrews;
>
> >>If the case is too difficult to understand, just consider a
> >>case of "vix.net" with just a single AAAA and no A glue
> >>returned. Then, IPv4 only host is at a loss.
>
> > No. It just asks the parent the missing glue records
> > individually. In this case for ?.gtld-servers.net/A records
> > from the root servers.
> >
> > The only time it will loose is when the NS RRset is too big
> > to even allow the requested glue RRset to be added
> > preferentially to the additional section.
>
> No. In general, clients can not know which nameserver needs
> glues and to which nameserver to ask. If a delegation loop
> involves multiple zones, clients have no reason to ask, for
> example, a "com." zone nameserver addresses of nameservers of
> zones under "edu."
The fully general case doesn't work even if the parents
return glue for out of scope servers because the caches
will reject it.
For the cross servers under the same parent the glue can
still be returned in the additional section even if it is
not a server for the zone it lives in.
> > Older parent servers that promote glue to answer don't even
> > have this problem.
>
> Glue is glue.
>
> Glue information should be answered only as glue.
I'm not suggesting otherwise. I was just pointing out that
the older servers did do this so lack of space would never
prevent the glue being returned.
> DNS works, if a child zone nameserver can be contacted at an
> address of a nameserver, which means most of the glue
> information can be incorrect.
>
> On the other hand, DNS answers are expected to be authoritative
> one or copy of it.
>
> Masataka Ohta
>
>
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: [EMAIL PROTECTED]
.
dnsop resources:_____________________________________________________
web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html
mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html
