Notify not having effect

[David Kirk]

> >The slave does receive the notify message but fails to update the zone
> >right away.
>
> "right away" ? but it does eventually transfer the zone or not?
>
> is the master answering authoritatively for the zone? if not, no axfr.
>
> can you manually transfer? on the slave:
>
> dig @master zone.com axfr
>
> >  Any idea why it's not getting updated? Is there some option I need to
set in
> >  the slave option field to tell it to listen to notifys?
>
> no.  in the zone statement, indicate the
>
> masters {ip_adress;};
>
> correctly.
>
> turn on logging on both machines and see what the complaints are.  I use
this:
>
> logging{
> channel my_file {file "/var/log/named.run"; severity debug; print-time
yes; };
> category default {my_file;};
> category panic {my_file;};
> category packet {my_file;};
> category eventlib {my_file;};
> category queries {my_file;};
> category lame-servers {null;};
> category cname {null;};
> };


How many other zones are being slaved on the slave server?

If I remember the NOTIFY behavior correctly, NOTIFY requests get
queued on a FIFO basis (I could be wrong).  If the name server is only
slaving a zone or two, these get dealt with quickly; as the number of
slaved zones go up, the more potential for delay when a master server
sends a NOTIFY to the slave server.

Another question ... if the other name server is under your explicit
control, do you have any means by which to operate in a dual-master
mode, i.e. have a central zone repository distribute out to both servers
the zones as they change?  That would eliminate dependency on
slaving the zones; of course, if the secondary server is not under your
direct control, this arrangement does not work.

Just some thoughts ...