NOTIFY-triggered Auto-slaving (was Re: how to list ALL zones of my master server)
Kevin Darcy
kcd at daimlerchrysler.com
Tue Oct 1 23:32:26 UTC 2002
Mark_Andrews at isc.org wrote:
> If you want this write it up as a draft and submit it to
> the IETF. The working group thought about this when we did
> NOTIFY but left it as a exercise for the future. At the
> time we were worried about security and theft of service
> in addition to the meta data problem. We also wanted the
> basic zone content updates to get through to RFC status and
> not to get bogged down in debate over what metadata needs
> to be tranmitted let alone how to do it.
>
> TSIG can provide the security and through that protection from
> theft of service. The rest is still a mess.
OK, Mark, try taking off your protocol-designer hat for a moment and put
on your BIND-implementor hat. Given that TSIG-signed NOTIFYs are now a
reality with BIND, without any protocol changes, RFCs or IETF/IESG action
required, and therefore security is no longer the issue it once was,
would it not be a reasonable feature request to have a BIND instance
automatically slave zones upon receipt of a TSIG-signature-verified
NOTIFY, if configured to do so by the administrator of said BIND instance
and using metadata provided, perhaps in template form, by said
administrator? It's a "halfway measure" only in the sense that we don't
have a whole protocol by which a master can communicate metadata to the
slave -- as you said, that's "still a mess" -- but just because we can't
do *everything* does that mean we should do *nothing* to automate
slave-creation? Is this an all or nothing proposition? I don't think so.
I know there a lot of DNS admins out there -- I'm one of them, since this
is one of the few things I haven't automated yet -- who want to eliminate
the drudge work of constantly editing named.conf files, and would eagerly
embrace a less-than-comprehensive solution to the problem.
- Kevin
More information about the bind-users
mailing list