Domain Name Registrar vs. Webcountry Hosting...Who do I believe?
Stefan Probst
stefan.probst at opticom.v-nam.net
Sun Nov 12 03:44:30 UTC 2000
At 20:15 09.11.00 +0200, Thor Kottelin wrote:
-------------------------
> > ls -d internethosting.co.za.
>[server.webcountry.net]
> internethosting.co.za. SOA www.webcountry.net
>admin.internethosting.co.za. (952720063 21600 3600 604800 86400)
> internethosting.co.za. NS www.webcountry.net
> internethosting.co.za. MX 10 mail.internethosting.co.za
> internethosting.co.za. A 216.65.111.4
> mail.internethosting.co.za. A 216.65.111.4
> www.internethosting.co.za. A 216.65.111.4
> internethosting.co.za. SOA www.webcountry.net
>admin.internethosting.co.za. (952720063 21600 3600 604800 86400)
>
> > ls -d internethosting.co.za.
>[ns2.webcountry.net]
> internethosting.co.za. SOA www.webcountry.net
>admin.internethosting.co.za. (952720063 21600 3600 604800 86400)
> internethosting.co.za. NS www.webcountry.net
> internethosting.co.za. MX 10 mail.internethosting.co.za
> internethosting.co.za. A 216.65.111.4
> mail.internethosting.co.za. A 216.65.111.4
> www.internethosting.co.za. A 216.65.111.4
> internethosting.co.za. SOA www.webcountry.net
>admin.internethosting.co.za. (952720063 21600 3600 604800 86400)
>
>There's only one NS record, and even that record points to a third server
>(216.65.111.5 instead of 216.65.111.2 or 216.65.111.3). At this point I'd
>probably switch to another provider...
My 0.02$:
The webcountry.net says that the domain is supposed to be delegated in the
parent's zone to the two name servers. Both of those name servers answer
authoritative. OK.
BUT:
The zone file itself doesn't declare that the zone is delegated to those
two name servers.
Technically this works. But, since records from the own zone "override"
the parent's delegation records, the records about the two nameserver
would in the resolver be overwritten with the single record about
"www....".
This renders the intended redundancy of the two name servers useless.
Why is it setup like this?
Oversight? A bit hard to believe, especially since they refuse to correct
it, i.e. replace the single NS record (which points not two www...) with
two NS records pointing to the two name servers.
Intention to fake two name servers, where there is in fact only one? Much
more probably. Especially since the used name server has obviously a "www
name", which doesn't sound very professional for a name server - and might
make the parent's hostmasters suspicious.
Concerning the SOA record:
It might be possible - and AFAIK ok - that they are using the www....
machine as "unpublished master", i.e. the original zone file would be
hosted on this www... machine, and the two name servers would be slaves to
that master. The zone would be delegated only to this two name servers
(not to the www... machine). But this delegation has to be done not only
in the parent's zone, but also via two NS records in the own zone file.
And this latter one is definetly missing.
HTH
Stefan
More information about the bind-users
mailing list