Eliot Lear writes:
your implementation
And others, notably BIND 8. We're talking about MOST of the Internet's
DNS servers here. (See http://cr.yp.to/surveys/dns1.html; I'm assuming
that *.com is representative.) This is not a small issue.
could cause interoperability problems by potentially allowing for
different contents of the same zone with the same serial number.
No. The _administrator_ can create problems by violating RFC 1034.
(Andrews has admitted that his broken configurations violate RFC 1034.)
The administrator cannot avoid breaking the rule. All they
can do is ensure that the time the rule is broken is
minimised.
If the administrator follows RFC 1034's consistency requirements, no
problems occur. In fact, if the administrator violates RFC 1034 but
still follows the easy semi-synchronization rule, no problems occur.
You don't live in the real world if you think there won't be
errors. The job of computers is to make life easier not
more complicated. Having a protocol/implementation that
doesn't deal with timing mistakes is a bad.
Andrews is finally coming out of the closet and arguing that the RFC
1034 rule should be changed (because, with his software, the rule is
hard for administrators to follow). But he can't reasonably argue
against the semi-synchronization rule. People normally follow that rule
anyway; there's no reason to break it; and it guarantees that problems
don't occur.
Dan if you want everyone to follow this rule then are you
going to modify your software to enforce it. You software
doesn't currently comply with this rule.
Mark
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET:
Mark(_dot_)Andrews(_at_)isc(_dot_)org