DNS server generates an enormous amount of traffic

[Irvine Short]

Oh, yes, I'm running BIND 4.9.7

Irvine Short <irvines at iafrica.com> wrote in message
news:7nl6t2$aqr$1 at nnrp01.ops.uunet.co.za...
> Hi again
>
> I narrowed the traffic down to named by two things:
>
> 1. If I killed named the traffic went down
>
> 2. a ps or top indicated that named was munching most of the CPU,
sometimes
> up to 100%.
>
> CPU is usually around 5%, if that.
>
> The primary name server is a HP Netserver EISA / PCI 100MHz with 72MB RAM,
> the secondary a P75 PCI system with 32MB RAM Neither have memory
problems -
> they're both configures with 128MB swap and never use more than about 5MB
of
> it according to top.
>
> The secondary is connected to the Internet via two  analog leased lines
> running multilink PPP, which is what most of the 5% is taken up with.
>
> The primary is on the 64k line via a router.
>
> I'm using FreeBSD 2.2.8, the primary RELEASE and the secondary STABLE
>
> I've been doing some reading, and tried a kill -USR1
>
> I've included it below.
>
> Are all those messages like:
>
> ns_resp: ns F.GTLD-SERVERS.NET rcnt 1 (busy)
> ns_resp: nsdata 207.159.77.18 rcnt 1 (busy)
> ns_resp: ns I.ROOT-SERVERS.NET rcnt 1 (busy)
> ns_resp: nsdata 192.36.148.17 rcnt 1 (busy)
>
> OK?
>
> I'm very new to name server debugging, they've always just worked for me.
>
> I did notice that if I pinged say www.ibm.net there was a lookup the first
> time, but the second time it answered it's own questions.
>
> Here follows some log file:
>