ISC BIND 9.3.5-P2 is now available

Mark Andrews Mark_Andrews at isc.org
Tue Aug 5 03:19:18 UTC 2008


> At Sat, 2 Aug 2008 11:21:47 -0600,
> The Doctor <doctor at doctor.nl2k.ab.ca> wrote:
> > 
> > Bug!!
> > 
> > I do not see this in the P1, but is plaguing the P2s:
> 
> > Aug  2 11:15:25 doctor named[14742]: starting BIND 9.3.5-P2
> > Aug  2 11:15:25 doctor named[14742]: found 1 CPU, using 1 worker thread
> > Aug  2 11:15:25 doctor named[14742]: loading configuration from '/etc/named
> .conf'
> > Aug  2 11:15:25 doctor named[14742]: listening on IPv4 interface em0, 204.2
> 09.81.1#53
> > Aug  2 11:15:25 doctor named[14742]: socket.c:485: unexpected error:
> > Aug  2 11:15:25 doctor named[14742]: fcntl(512, F_SETFL, -1): Bad file desc
> riptor
> 
> Try specifying a small value such as 32 for the reserved-sockets
> option:
> 
> options {
> 	reserved-sockets 32;
> 	...
> };
> 
> But, even if this seemingly solves this error, I suspect the above
> error indicates that your OS has a fundamental limitation on the
> number of file descriptors and will soon trigger a different type of
> trouble.  If you've not done this yet, I'd suggest you check the OS
> capability with the tool I posted a few days ago, which is available
> at http://www.jinmei.org/selecttest.tgz
> 
> One more thing: as already pointed out, if you specify a specific port
> using query-source:
> 
> > Aug  2 11:15:25 doctor named[14742]: /etc/named.conf:60: using specific que
> ry-source port suppresses port randomization and can be insecure.
> 
> you effectively disable any security feature of P1 or P2 or recent
> beta versions.  Using these versions with a specific port doesn't
> provide anything new and is meaningless.
> 
> ---
> JINMEI, Tatuya
> Internet Systems Consortium, Inc.

	What's wierd here is that fcntl(F_DUPFD) succeeded but
	fcntl(F_SETFL) then failed on the value returned from
	fcntl(F_DUPFD).

	Mark
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark_Andrews at isc.org


More information about the bind-users mailing list