doesn't look like it (looking @ tcpdump anyway). looks like when named
on the parent domain starts up and gets queried for the child domain, it
queries all the NS in the list, then sticks to one (so all subsequent
queries to the parent for the child go to the same child NS). My ques
is: when does the parent start querying different child NS?
Mark_Andrews@xxxxxxx wrote:
>
> Whenever the server get queried. The servers for the zone
> not queried get there estimates improved so that they will
> eventually be tried again.
>
> > one last ques. on this topic: anyone know how often does the RTT info
> > get refreshed?
> >
> > Anthony Golia wrote:
> > >
> > > OK I wasn't doing anything wrong, the foo.com srvr just queried all the
> > > ad.foo.com srvrs in the list, one per query that *it* got. I thought it
> > > would have queried *all* in the list (for, say the SOA rec) when one
> > > query came into it, learned the RTTs all at once, then queried the
> > > closest ad.foo.com srvr. that way it would never query one far away for
> > > the actual data...
> > >
> > > Anthony Golia wrote:
> > > >
> > > > Thx, that's how I thought it worked but mine seems to be round
> > > > robining. I am doing something wrong. tcpdump shows that the named
> > > > never queries all four (for RTTs or anything) but just queries one
> > > > (whenever i try to look something up in ad.foo.com) ad NS srvr at a time
> > > > in a RR fashion.
> > > >
> > > > from foo.com:
> > > >
> > > > ad IN NS nyad1
> > > > ad IN NS nyad2
> > > > ad IN NS ccad1
> > > > ad IN NS lnad1
> > > >
> > > > from named.conf:
> > > >
> > > > options {
> > > > directory "/var/named";
> > > > pid-file "/etc/named.pid";
> > > > check-names master warn;
> > > > check-names slave warn;
> > > > check-names response ignore;
> > > > forwarders {
> > > > 199.89.110.57;
> > > > 199.89.99.57;
> > > > 199.89.110.56;
> > > > 199.89.99.56;
> > > > };
> > > > forward only;
> > > > };
> > > > .
> > > > .
> > > > .
> > > > zone "foo.com" {
> > > > type master;
> > > > file "zone/foo.com";
> > > > forwarders {};
> > > > };
> > > >
> > > > Pete Ehlke wrote:
> > > > >
> > > > > On Wed, Feb 26, 2003 at 04:02:03PM -0500, Anthony Golia wrote:
> > > > > > I mean for subdomains. It seems to round robin among them. Is that
> > how
> > > > > > it works? is there a way to get it to use the first one if it's up,
> > but
> > > > > > the next if it's not, etc.? Any help is appreciated.
> > > > > >
> > > > > Glossing over the details substantially:
> > > > >
> > > > > BIND keeps a running scorecard of which server responds fastest, and
> > > > > prefers that one. The scorecard is recalibrated periodically.
> > > >
> > > > --
> > > > NOTICE: If received in error, please destroy and notify sender. Sender
> > > > does not waive confidentiality or privilege, and use is prohibited.
> > >
> > > --
> > > NOTICE: If received in error, please destroy and notify sender. Sender
> > > does not waive confidentiality or privilege, and use is prohibited.
> >
> > --
> > NOTICE: If received in error, please destroy and notify sender. Sender
> > does not waive confidentiality or privilege, and use is prohibited.
> >
> --
> Mark Andrews, Internet Software Consortium
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE: +61 2 9871 4742 INTERNET: Mark.Andrews@xxxxxxx
--
NOTICE: If received in error, please destroy and notify sender. Sender
does not waive confidentiality or privilege, and use is prohibited.
|
