bind 9 chrooted & syslogd

Mark_Andrews at isc.org Mark_Andrews at isc.org
Tue Oct 7 05:20:13 UTC 2003


> Hello,
> I have bind 9.22 chrooted on SuSE-Linux 8.2 on a sysloghost, that is
> receiving from several other machnines. 
> The problem: Whenever I restrart named, the process just slows down and does
> not answer up anymore. It takes roughly ninety minutes just to start up and
> load all the zones. Even then it does not answer queries, though the
> processes are alive. 
> I poked around and found that syslogd needs to be restarted and the problem
> is handled. 
> The question: Is this a) normal behaviour ? b) a bug in syslogd c) a bug in
> bind d) something else ? 
> 
> Syslogd is running with:
> /sbin/syslogd -r  -a /var/lib/dhcp/dev/log
> 
> Has anyone encountered/handled this w/o restarting syslogd all the time ? 
> 
> Thanks for reading
> 
> Dan

	This sounds like named and syslogd are in a deadlock.

	syslog writes in named are blocking because syslogd is
	stalled on reverse lookups which require named to answer.

	I generally just stop syslogd performing the reverse lookups.
	I haven't bothered to check if the SuSE-Linux syslogd has such
	option.

	Killing syslogd allows the syslog() calls to fail and named
	to proceed.

	Other things that may help are changing the service order of
	hostname lookups to use files first and ensuring /etc/hosts
	is fully populated for all hosts syslogging to this box.
	Use a packet filter to block syslog messages from machines
	not listed in /etc/hosts.

	Mark
--
Mark Andrews, Internet Software Consortium
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