On Fri, 25 May 2001, tc lewis wrote:
>
> i understand you don't mean legally as in i go to jail.
>
> my question is why do the registrars require a 1 to 1 mapping (if they
> even do -- if not, why does opensrs)?
>
That's what I thought your question was (notwithstanding all the thread
drift about needing or not needing two different nameservers)
I think, it's because in .com/.net/.org (NSI/Verisign registry) they put
glue records in the roots for each nameserver. If you had multiple
A recs for a given IP address, when it came time to change it, it would
require an exhaustive search of all the glue records in the root zone,
everytime.
This way, you don't.
Also, it is not necessarily a 1-to-1 rule, you can legally go the other
way, round-robining a single nameserver record over multiple IP addresses.
> the root servers provide A records for registered nameservers, but not PTR
> records. having 2 A records to the same ip works fine.
>
> an example is cr.yp.to's nameservers:
> additional: a.ns.yp.to 235171 A 131.193.178.181
> additional: b.ns.yp.to 235171 A 131.193.178.181
>
Perhaps the .to roots do not do this. Many ccTLD's don't (they just let
the nameservers resolve via DNS); .CA also does not do this, so you
might be able to get away with having multiple hostnames on a single
IP address serving as nameservers (until you try using one as a nameserver
for a .COM/.NET/.ORG domain)
-mark
--
mark jeftovic
http://www.easydns.com
http://mark.jeftovic.net
