Re: Problems moving to 9.3.0

>>>>> "John" == John Klos <john@ziaspace.com> writes:

John> Moving from BIND 9.2.3 to 9.3.0, I've been seeing some
John> problems. On a PowerPC NetBSD 2.0 system on which 9.2.3
John> works perfectly, I get this from 9.3.0:

John> Oct 20 17:04:54 lain named[368]: rbt.c:1584: INSIST(child != (void *)0) failed
John> Oct 20 17:04:54 lain named[368]: exiting (due to assertion failure)

Assertion failures means that BIND9 has detected an internal
inconsistency. You should file a bug report.

John> So what is so special about ziaspace.com as opposed to
John> ziaspace.nl?

Well the two zones have different NS RRsets.

John> The other problem I'm having is with a NetBSD 1.6.2 machine
John> and BIND 9.3.0. When I launch it, I see this in the logs:

John> Oct 20 11:24:56 grittykitty named[19879]: loading configuration: unexpected end of input
John> Oct 20 11:24:56 grittykitty named[19879]: exiting (due to
John> fatal error)

John> The named.conf on that machine has run fine with 9.2.3 for
John> ages, and it's identical to the configuration files on three
John> other machines, including another 1.6.2 PowerPC
John> machine. named-checkconf reports nothing.

John> So my question is this: if named-checkconf doesn't report
John> anything, and I can run the same named.conf elsewhere, then
John> how to do I get more detailed a report of BIND's problem so
John> I can see what going on? I corrected the problem by removing
John> some excess whitespace, but it's not helpful that
John> named-checkconf doesn't have the same rules for what will
John> and what won't work as BIND. Ideas?

named-checkconf and named use the *same code* to parse named.conf files.
Perhaps you're using different version of named-checkconf from the one
that came with the BIND9 distribution? And are you sure the named.conf
files really are identical?