dynamic update transfers to slave servers

[/dev/rob0]

On Monday 13 June 2005 15:51, Trey Nolen wrote [offlist]:
> > Updates sometimes take as much as 2 seconds to get to all the
> > slaves, when network lag is bad.
>
> Wow..that would be great....I'd be happy with a few minutes, but 2
> seconds would really be nice...
>
> > I don't know BIND 8, but I can tell you that BIND 9 is good at
> > this.
>
> I had tried going to BIND 9 a while back (maybe 2 years ago..hard to
> keep track), but I had a lot of problems with it just not taking
> requests after being up a few days.  That causes all kinds of havoc,

I suggest that you put up your dynamic zone on a BIND 9 master and 
slaves, separate from your main zone. (If you need records in the main 
zone for your dynamic clients, use CNAMEs.) Then you can experiment 
with BIND 9 again, first as a slave to see how it goes.

> > I would be interested to compare notes on your implementation.
> > Mine's powered by a Perl CGI frontend to a shell script which runs
>
> This is also a set of perl scripts that calls nsupdate. I am using a
> package called GnuDIP.  Here is the webpage: 
> http://gnudip2.sourceforge.net/

Thanks for the pointer. Wow, registered in 2001, but I didn't find it 
when I was looking around to set up my own service, so I reinvented a 
wheel ...
-- 
    mail to this address is discarded unless "/dev/rob0"
    or "not-spam" is in Subject: header