one short question
Kevin Darcy
kcd at daimlerchrysler.com
Sat Jun 2 01:50:07 UTC 2001
Brad Knowles wrote:
> At 8:49 PM -0400 6/1/01, Kevin Darcy wrote:
>
> > Well, my post was mostly in jest,
>
> As was my original response. Actually, I got a bit of a kick
> throwing around numbers like that, and being reasonably confident
> that I actually got them right.
>
> > but just for record, the original poster
> > did, after all, specifically ask how to get all of the A records in a zone
> > (thereby rendering irrelevant CNAMEs and IPv6 address records).
>
> Technically, the original question was "how can I query all the
> IN A which are in a zone", but since this doesn't quite make sense,
> we all interpret (or misinterpret, perhaps with glee ;-) this
> question in our own way.
Point well taken. He did ask how to *query* (rather than get or obtain or
determine) all A records in the zone. So my answer is technically wrong, and yours
is right. One would have to query all *possible* legal names in the zone.
BTW, shouldn't dot/period be included in the character set? This complicates the
calculation somewhat, since of course you can't have an empty label or one which
starts or ends with a dash...
> > Multiple
> > A records can point to the same address, true, but my (facetious) assumption
> > was that every A record had a corresponding PTR record:
>
> Do you mean that multiple names could point to the same A record?
> If so, yes that is certainly true.
>
> However, it would be perfectly legal to have multiple PTR records
> per IP address, and that could make things really nasty.
>
> > As for RFC 1918 or not-yet-assigned
> > ranges, I don't think we can rule those out, since the poster didn't specify
> > that this was an Internet zone -- internally folks might use RFC 1918
> > addresses and/or addresses from not-yet-assigned ranges.
>
> Considering that he said that zone transfer didn't work, I think
> we can safely assume that he probably doesn't have to worry about RFC
> 1918 addresses.
>
> > I think
> >you're right
> > about Class D, though...
>
> And thinking about it again, simply generating all possible IP
> addresses within the specified range is probably more along the lines
> of what he was thinking about.
>
> But, what if there were multiple IP address ranges (for the
> various names in the particular zone), some of which he may not even
> know? What if some of those names are not CNAME records, but are
> also not the true canonical name for the machine in question (i.e.,
> you have multiple names resolving to the same IP address, but not
> through CNAME chains)?
It boggles the mind.
- Kevin
More information about the bind-users
mailing list