I first reported this issue on 7/23 (Windows 2003, BIND 9.4.2-P1) so far
with no acknowledgement from anyone at ISC. Seems Windows BIND is
perhaps a "redheaded step-child". I am grateful to Vinny Abello and
Radim Hajek for their posts concerning the same problem. I am still not
clear on whether this is a BIND flaw or Windows DLL flaw.

Anyway, I have discovered that the named.exe crash produces several text
and binary files (.hdmp and .bdmp) which are immediately erased if one
closed out of the crash message. I successfully saved a couple sets of
these files and could provide them to ISC if they would be useful.


Bob Warren
Semiconductor Research Corp.

-----Original Message-----
From: bind-users-bounce@isc.org [mailto:bind-users-bounce@isc.org] On
Behalf Of Vinny Abello
Sent: Saturday, July 26, 2008 11:24 PM
To: Radim Hajek; bind-users@isc.org
Subject: RE: Bind 9.5.0-P1 WinNT version - Crashes

There is currently no patched working version of BIND available for the
Windows platform from my own tests and many others who have responded to
me. 9.4.2-P1, 9.5.0-P1 and the 9.5.1 betas all seem to have issues with
stability or memory utilization that makes them unstable. 9.4.2 is
perfectly stable in our environment. We're contemplating switching to
FreeBSD for the OS on our recursive name servers.


> -----Original Message-----
> From: bind-users-bounce@isc.org [mailto:bind-users-bounce@isc.org] On
> Behalf Of Radim Hajek
> Sent: Saturday, July 26, 2008 4:41 AM
> To: bind-users@isc.org
> Subject: Bind 9.5.0-P1 WinNT version - Crashes
> Hello,
> after upgrading bind to last 9.5.0-P1 on one of our Windows 2003
> server BIND starts crashing several times a day.
> Faulting application named.exe, version, faulting module
> ntdll.dll, version 5.2.3790.3959, fault address 0x0002a36b.
> There are NO errors in bind log, except one, BUT it doesnt correspond
> to other crashes...
> 25-Jul-2008 16:04:28.864 critical: ..\mem.c:1716: INSIST(mpctx-
> >freecount > 0)

> failed
> 25-Jul-2008 16:04:28.864 critical: exiting (due to assertion failure)
> I have found another messages from users who have the same problem in
> another lists - but no answers.
> Thanks
> Radim Hajek