Bind 8.1.2 / Solaris 2.5.1 / zone transfert timeout

[Mark_Andrews at isc.org]

	You will have installed security patches which have overwitten
	named-xfer from the BIND distribution.  Re-install named-xfer.

	Mark

> Hi,
> 
> 
> i've got a problem, with one of my dns server :
> 
> 
> My primary DNS (v 8.1.2) server (Solaris 2.5.1) is also secondary for
> other sub-domain (to collect delegation data).
> The syslogs says :
> Aug 20 13:58:16 alpinia2 named[7788]: zone transfer timeout for
> "fina.be"; pid 6577 killed
> Aug 20 13:58:16 alpinia2 named[7788]: named-xfer "fina.be" exited with
> signal 15
> Aug 20 15:50:51 alpinia2 named[7788]: secondary zone "fina.be" expired
> Aug 20 16:13:31 alpinia2 named[7788]: secondary zone "fina.be" expired
> 
> 
> and the process stay in memory :
> alpinia2. # ps -ef | grep named
>     root 17598  7788 23 16:01:56 ?       20:13 /usr/sbin/named-xfer -z
> upstream.total.com -f secondary/db.upstream.total.com -
>     root 17597  7788 18 16:01:56 ?       19:35 /usr/sbin/named-xfer -z
> 224.191.249.146.in-addr.arpa -f secondary/db.224.191.24
>     root 20282  7788 23 17:02:00 ?        5:42 /usr/sbin/named-xfer -z
> 160.191.249.146.in-addr.arpa -f secondary/db.160.191.24
>     root  7788     1  3   Aug 10 ?       129:09 /usr/sbin/named -b
> /opt/bind/v812/config/named.conf
>     root 17082  7788 17 15:50:51 ?       20:20 /usr/sbin/named-xfer -z
> fina.be -f secondary/db.fina.be -s 0 -C 1 -P 53 193.74.
> 
> 
> I can connect to theses machine, and the named daemon run, because
> other subdomain works.
> My secondary server works perfectely, same OS & Bind version. 
> The named.conf doesn't be modified. 
> Below you'll see for some sub-domain the process begin by closing some
> file descriptors from 64 to 3.
> But for sub-domain which doesn't work it begin at 2^31 - 1  and then
> he fells into time-out later.
> 
> When it's working :
> ..
> 774:	sysconfig(_CONFIG_OPEN_FILES)			= 64
> 774:	close(63)					Err#9 EBADF
> 774:	close(62)					Err#9 EBADF
> 774:	close(61)					Err#9 EBADF
> 774:	close(60)					Err#9 EBADF
> 774:	close(59)					Err#9 EBADF
> 774:	close(58)					Err#9 EBADF
> ...
> 774:	close(7)					Err#9 EBADF
> 774:	close(6)					Err#9 EBADF
> 774:	close(5)					Err#9 EBADF
> 774:	close(4)					= 0
> 774:	close(3)					= 0
> 774:	brk(0x000C4F18)					= 0
> 774:	brk(0x000C6F18)					= 0
> 774:	open("/dev/conslog", O_WRONLY)			= 3
> ..
> 
> When it's not working (timeout)
> 777:	sysconfig(_CONFIG_OPEN_FILES)			= 2147483647
> 777:	close(2147483646)				Err#9 EBADF
> 777:	close(2147483645)				Err#9 EBADF
> 777:	close(2147483644)				Err#9 EBADF
> 777:	close(2147483643)				Err#9 EBADF
> ..
> 777:	close(2147483027)				Err#9 EBADF
> 777:	close(2147483026)				Err#9 EBADF
> 777:	close(2147483025)				Err#9 EBADF
> 775:	vfork()						= 779
> 779:	vfork()		(returning as child ...)	= 775
> 779:	execve("/usr/sbin/named-xfer", 0xEFFFDD18, 0xEFFFFCF4)  argc =
> 13
> 
> And finally, when I launch the command named-xfer manually, the zone
> is transfered ! 
> 
> 
> Then the questions are :
> --> Where does it come from ?
> --> What can I do to fix this problem, without extending the time-out
> duration ?
> 
> Thanks a lot.
> 
> Followup-To comp.ptotocols.dns.bind
> 
> 
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org