4 second delay on lookups
Gavin Hamill
gdh at acentral.co.uk
Fri Jul 7 17:51:07 UTC 2006
Hi :)
I'm running BIND 9.2.4 as provided by Debian in the Sarge distribution,
and am finally getting round to looking at a problem that has bugged
me for months.
A tcpdump will easily tell the story. Client PC is 10.0.0.29, gateway
is 10.0.0.1 on the LAN and 80.192.144.33 on the WAN.
Gateway machine, LAN interface:
18:03:01.962740 IP 10.0.0.29.32782 > 10.0.0.1.53: 38294+ A? sun.com.(25)
18:03:06.284149 IP 10.0.0.1.53 > 10.0.0.29.32782: 38294 1/4/0 A
72.5.124.61 (113)
Gateway machine, WAN interface:
18:03:05.966780 IP 80.192.144.33.32768 > 192.52.178.30.53: 31993 A?
sun.com. (25)
18:03:05.999320 IP 192.52.178.30.53 > 80.192.144.33.32768: 31993-
0/4/4 (161)
18:03:06.001387 IP 80.192.144.33.32768 > 192.18.99.5.53: 28178% [1au]
A? sun.com. (36)
18:03:06.138290 IP 192.18.99.5.53 > 80.192.144.33.32768: 28178 FormErr
[0q] 0/0/0 (12)
18:03:06.139178 IP 80.192.144.33.32768 > 192.18.99.5.53: 56235 A?
sun.com. (25)
18:03:06.282613 IP 192.18.99.5.53 > 80.192.144.33.32768: 56235*- 1/4/4
A 72.5.124.61 (177)
There is exactly a 4-second wait between BIND receiving a request and
asking a root-server where it can find NS for sun.com - why? :)
A cursory search has brought up
http://www.redhat.com/archives/nahant-list/2005-October/msg00375.html
but I don't wish to disable IPv6 (even though I'm not using it at
present) - what can I do to get more information and eliminate this
delay?
My named.conf is trivial + stock-Debian.. nothing in options {} except
the directory "/var/cache/bind";, no logging {} and only root,
localhost, (127, 0, 255) in-addr.arpa zones...
Cheers,
Gavin.
More information about the bind-users
mailing list