non-recursive request to forwarders

[Simon Waters]

Valentin Nechayev wrote:
> 
> I don't want to specify forwarders which aren't public servers of the zone.

Argh - got it - sorry being a bit slow.

I thought you were trying to sneak around bad connectivity, by
forwarding to the recursive servers of your peers........ which
would have been underhand, but you aren't.

Hmm maybe you could use "stub", and hardwire key DNS servers
address be adding a fake domain for that server.

i.e. zone "ua" ; type stub
     zone "ns.lucky.com" ; type master

It looks crazy and difficult to maintain, but it might work,
never tried it myself.

With some twisted thinking (multiple A records for the fake
zones) you might include private slaves of a zone, but that
would be madness, as the lameness algorithmn would get in the
way (I think).

A different approach would be to secondary the "." zone, it
isn't large, but unless you have relevant TLD servers local to
you that may not help much, still it is one layer down.

> Second, it isn't correct to keep slave copy.

?!

In most cases where you aren't a parent, I'd have thought slave
was preferable to stub. I'm cynical about stub zones, I once
found a genuine use for them as part of a network renumbering
plan, but it was more functional than elegant!

> > Oh and a UA oddity, ns.kolo.net doesn't offer recursion, but
> > still returns additional data from cache, given it appears to
> > have a fair chunk of data cached, probably a bad move, but not
> > as bad as offering recursion.
> 
> Named replies with its cache contents applicable to the query even in case
> of non-recursive answer (both with recursion denied or not requested).

It is optional in named these days I believe, cached data
shouldn't be returned by TLD servers even if there isn't an easy
feature of the software.

> And it has *nothing* common with my question.

I was just looking for problems in UA domain that might be
making matters worse, since your request is fairly unusual.