Hi Sylvain,
> [beuc - Thu Oct 15 14:37:56 2009]:
> When 140.186.70.68 sends a notification to ns1.gnu.org about the
> savannah.gnu.org zone, something weird happen, because ns1.gnu.org is
> usually updated after ns2.gnu.org, while ns2 is the slave.
>
> I test with:
> dig savannah.gnu.org SOA @ns1.gnu.org
> dig savannah.gnu.org SOA @ns2.gnu.org
>
> So apparently ns1 relays the notification, but doesn't update
> immediately. It also takes longer than the TTL (300) for it to
> refresh.
>
> Do you know how this works?
Yeah, this was caused by Bind being rather silly when one uses views +
notifies. We have 2 views; your notify arrived in the trusted view
(because it originated from one of our IPs), and that view actually did
get updated, and sent out notifies to the other nameservers which
updated properly.
Turns out Bind does not also update the cached zone file in the other
views without some extra configuration to make the trusted view notify
the other view... Sigh, this is *so* much simpler with djbdns. Anyway,
it's fixed now.
Thanks,
Ward.
--
Ward Vandewege <
ward@...>
Free Software Foundation - Senior System Administrator
