ixfr problem

[Bill.Stephens at fritolay.com]

After testing I found out I was sadly mistaken.  As usual Cricket was absolutely
correct, after all he (and Paul Albitz) wrote the book.  This obviously makes my
life a lot easier.

Thanks




Kevin Darcy <kcd at daimlerchrysler.com>
10/08/2001 05:29 PM

To:   BIND-Users <bind-users at isc.org>
cc:
Subject:  Re: ixfr problem


Cricket Liu wrote:

> > Win2k servers on remote subnets are set up to point to their local slave
> DNS
> > server first then the master server second.  The domain controllers update
> their
> > records every 20 minutes or so (we plan to decrease the frequency).  When
> they
> > send their updates, they go to the slave, which forwards the update to the
> > master.  The master updates immediately, the slave appears to get updated
> within
> > 30 minutes.  IXFR is involved because we're just using IXFR, rather than
> the
> > full AXFR to update the slaves.
>
> You sure about that?  The Windows 2000 dynamic update routines send their
> updates to the name server listed in the MNAME field of the SOA record of
> the zone they need to update, not to the first name server listed in their
> resolver's
> configuration.

Some things I've heard from our local Microsoft techs implies that, at least
under some circumstances, Win2K will send updates to the "preferred
nameserver", i.e. the first in the resolver list. Presumably (hopefully) it
then falls back to using the SOA.MNAME nameserver if that doesn't work. This
behavior probably depends a lot on how you configure the client...


- Kevin