[dns-operations] Client retry behavour?

[Shumon Huque]

On Wed, May 09, 2007 at 12:52:51PM +0200, Stephane Bortzmeyer wrote:
> On Wed, May 09, 2007 at 10:25:08PM +1200,
>  Simon Lyall <simon at darkmere.gen.nz> wrote 
>  a message of 55 lines which said:
> 
> > I was worrying about my recursive name servers [...] The driver
> > behind this was did I need to make the primary extremely reliable or
> > would clients ( most home users) just use the secondary (and
> > tertiary) name servers OK?
> 
> My experience on Unix is that the stub resolver included in the libc
> takes a long time (several seconds, which is typically unbearable) to
> timeout and to move to the second name server listed in
> /etc/resolv.conf. Worse, since it has no memory, the problem appears
> each time you run a new program.

The timeout parameter can be tweaked (at least on resolvers based
on bind 8.x and later). I usually put the following in resolv.conf
on the UNIX machines:

options timeout:1

I think modern windows resolvers fail over in 1 second also.

> This certainly makes a case for having a very reliable name server, by
> using clustering or anycasting. (Or to install a full resolver on any
> Unix machine.)

One option for making a more reliable primary resolver, is to deploy
multiple servers using the same anycast address.

--Shumon.