Variations on lame delegations (terminology question)

[Mark_Andrews at isc.org]

> 
> Mark.Andrews at isc.org wrote:
> >>>>>	They are all lame delegations.
> >>>>
> >>>>hmm but for some reasons, isc bind lame-ttl option has a different 
> >>>>opinion, and considers server to be lame only in the first case :-(,
> >>>
> >>>
> >>>	Well the second case you can't tell the difference between
> >>>	that and a nameserver that is down which is how it is
> >>>	treated.
> >>
> >>do i really need to know?, what is the point using incredible system 
> >>resources of a bind recursive server, just to follow up with servers, 
> >>which are apperantly down, and everybody except the bind itself has 
> >>figured it out.
> > 
> > 
> > 	Named penalises the rtt estimates of nameservers that are down.
> 
> but it doesn't stop bind to use them and use them and use them again and 
> again, if there is no better choice, thus causing very interesting 
> moments for many recursive server administrators? Is it really bad idea 
> to have lame-ttl covering the second and case as well, if we all agree 
> this is lame as well.
> 
> Ladislav

	If your link to the rest of the world is down then comes back
	up.  How long do you want named to wait before it starts to
	use the link again?

	Named can't tell the difference between a link down, host
	down, no server running, firewall blocking the packet,
	lots of packet loss.

	Mark
--
Mark Andrews, Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews at isc.org