Due to circumstances beyond my control I cannot use
BIND 9 until it supports the "fixed" rrset-order.
Until that time BIND 9 is useless to me. In fact,
it's worse than useless, because another
administrative group insists on running BIND 9 and
it doesn't play well with BIND 8.

I set up my master BIND 8 server with two network
interfaces. The slaves are configured to use IXFR
on the primary interface, and AXFR on the secondary
interface. That way if the IXFR fails AXFR is used
immediately and the zone is transferred. I
implemented this several years ago and hadn't had
a problem until BIND 9 was brought into the mix.

BIND 9.2.5 seems to work with BIND 8, but earlier
versions don't. It's probably because of this
Bug-fix:

1722. [bug] Don't commit the journal on
malformed ixfr streams.
[RT #12519]


ISC's download page for 9.2.5 (released March 2005)
shows about 30 upcoming fixes. The download page
for 8.4.6 (released January 2005) shows about 20
upcoming fixes. That doesn't look like dead code
to me.

-----Original Message-----
From: bind-users-bounce@isc.org [mailto:bind-users-bounce@isc.org] On =
Behalf Of Jim Reid
Sent: Friday, May 06, 2005 3:19 AM
To: Mike Mitchell
Cc: Brad Knowles; bind-users@isc.org
Subject: Re: BIND 8.2.7 master ixfr to 9.2.2 slave=20

>>>>> "Mike" =3D=3D Mike Mitchell writes:


Mike> Why is the answer "don't do that" instead of "hmm.. Lets see
Mike> if we can fix the bug"?

Because BIND8 is pretty much dead code.