short ttl and servermmirroring

[Kevin Darcy]

If the clients were written properly, you could just configure all of the
authoritative servers to give out the relevant addresses in "fixed"
primary/backup order and then the clients would fall back almost
*instantly* without the need for any resource-crunching
"superfast" master/slave convergence.


- Kevin

hilgart at my-deja.com wrote:

> At the suggestion of Jim Reid, I looked at the code and found that BIND
> does indeed intentionally put a delay in sending out NOTIFY from
> anywhere between 5 seconds to 15 minutes, using a random value.  The
> delay is roughly proportional to the number of zones you have.
>
> For us this is unfortunate as this is an ideal high-availability
> mechanism.  It looks simple enough to re-write the code, and I may do
> so.  Once we sign up for a support contract with the ISC, I will suggest
> that some more switches be offered around this cool NOTIFY feature.
>
> I'm surprised there hasn't been more request for this.  If you want to
> switch the IP of your Web server in seconds (or at least within
> minutes), you need 1) a short TTL on its record and 2) all authoritative
> servers to have that new IP ASAP .?  This thread is focussed on 2) .
> Isn't that how Distribued Director and Resonate Global Dispatch work?
>
> Or am I missing something?
> -john
> BASF Corporation
> In article <FAzq4.50$M31.3128 at burlma1-snr2>,
>   Barry Margolin <barmar at bbnplanet.com> wrote:
> > In article <88citf$36h$1 at nnrp1.deja.com>,  <hilgart at netscape.net>
> wrote:
> > >I have a related question.  I'm really eager to use the BIND 8 NOTIFY
> > >FEATURE but in my tests between to BIND 8.2.2 p 5 servers, it's
> taking
> > >longer than I'd like - anywhere from 30 seconds to a few minutes.
> That
> > >is, the primary server is waiting that long before initiating a
> NOTIFY
> > >to the secondary when a domain has changed.  The secondary picks up
> the
> > >new zone instantly once it has been NOTIFY'd.
> > >
> > >I was hoping for a 1- or 2-second value for this feature to kick in.
> > >These timings are key to enable rapid failover to back-up servers.
> > >
> > >What are other people's experiences?  What might be slowing this
> down?
> >
> > I think this is intentional.  It waits a random amount of time to keep
> from
> > flooding a slave server with NOTIFY messages all at once.
> >
> > The purpose of NOTIFY isn't to achieve instant propagation, just to be
> much
> > better than waiting for the Refresh period.
> >
> > --
> > Barry Margolin, barmar at bbnplanet.com
> > GTE Internetworking, Powered by BBN, Burlington, MA
> > *** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to
> newsgroups.
> > Please DON'T copy followups to me -- I'll assume it wasn't posted to
> the group.
> >
> >
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.