dns lookup timing cycling up and down

[Jason Vas Dias]

On Tuesday 31 January 2006 15:34, rsesnaski at yahoo.com wrote:
>  I've got a RHEL4 ES Update1 host running bind 9.2.4. My DNS lookup
>  times seem to be cycling up and down. Everything is fine for awhile
>  and then everything slows down and I start getting timeouts. I'm not
>  asking for the solution but does anyone know what areas to start
>  looking in or how to turn on some logging where I can get lookup
>  times. I'm no bind guru and this thing is driving me nuts.
>  
>  I'm running it in a chroot'ed /var/named/chroot/var/named directory,
>  how
>  easy is this to turn off, just copy the files up to /var/named and turn
>  off the chroot option in /etc/sysconfig/named. I think running it in a
>  chroot environment doesn't allow me to hook to it with an strace -p.
>  
>  Ron
>  
Hi -

Sorry for the problems you are experiencing.

I'm guessing you're being hit by this bug:

   1773.   [bug]           Fast retry on host / net unreachable. [RT #13153]

Basically, whenever an ENETUNREACH / EHOSTUNREACH error occurred on a query
packet send, the server would time out for 2 seconds (and retry twice). This
could happen if there is no route to a nameserver, eg. if an ns lookup returns
an IPv6 address and you have no IPv6 network routes.
 
This is fixed with the current BIND 9.3.0+9.2.5+ releases, and has been 
backported to the next release of bind for RHEL-4: bind-9.2.4-16.EL4 .
This release is currenly undergoing Red Hat QA testing for the next
RHEL-4 update release. Meanwhile, it can be downloaded from:
  http://people.redhat.com/~jvdias/bind/RHEL-4/9.2.4-16.EL4
There is also BIND 9.3.2 compiled for RHEL-4, available from:
  http://people.redhat.com/~jvdias/bind/RHEL-4/9.3.2-1_EL4

Perhaps you might consider giving one of these releases a try
and see if it solves your problem - if not, please let me know.

Thanks & Regards,
Jason Vas Dias
bind package maintainer
Red Hat, Inc.