wael.shahin at gmail.com
Thu Sep 28 19:50:11 UTC 2006
I believe that what Dave have said is true, am sure that will generate more
traffic on my servers but it will be the negative responses only. I have
been monitoring the servers since then and they are OK, I actually don't
know if the -4 argument would be a better solution since i haven't tried it.
and also, we on peak time have almost 20'000 concurrent users and servers
utilization is ok though.
On 9/28/06, Joseph S D Yao <jsdy at center.osis.gov> wrote:
> On Thu, Sep 28, 2006 at 08:58:36AM -0700, Dave Stewart wrote:
> > On Sep 27, 2006, at 5:13 PM, Joseph S D Yao wrote:
> > >
> > > On Wed, Sep 27, 2006 at 03:12:49PM -0700, broadcast wrote:
> > > ...
> > >> Well, I havent used the -4 option, everything seems to be working
> > >> properly with the ncache 1 second option.
> > >
> > >
> > > Ah. Well this is useful feedback. Thanks.
> > Broadcast had mentioned the ncache setting resolved the problem in a
> > previous message, I figured that implied he hadn't tried the -4
> > option (people tend to stop trying things when they stumble upon
> > something that works:). Maybe you missed that Joe.
> Yes, I went back and looked, and that was one of two messages in this
> thread I had inadvertantly skipped.
> > My question as a student of Bind and not an expert is this: is there
> > any "anti-social" behavior occurring because of the ncache being set
> > to such a short time? It would seem to me that this might cause
> > Broadcast's bind server to repeatedly query these bad domains (since
> > the negative response isn't being cached, therefore the domain isn't
> > in the cache ...), increasing traffic. If so, I would guess this can
> > be mitigated against by using the -4 option, reducing DNS noise on
> > Broadcast's servers. Or am I missing something or otherwise making a
> > mountain out of a molehill?
> Yes, you are right, which is why this isn't normally set so low. But
> perhaps his DNS server does a small number of queries relative to those
> in the world. And for those who missed it (I know you didn't), positive
> answers should still be cached for the full length of time.
> Joe Yao
> This message is not an official statement of OSIS Center policies.
Dry Networks don't pass by lakes !!!
More information about the bind-users