Zone Transfer Issues With BIND9

[Kevin Darcy]

Nate D. wrote:

>On Oct 30, 2003, I posted a message about having zone transfer issues
>between two servers, both with an internal and external view. I made
>sure I accurately specified the match-clients, notify-source,
>transfer-source, and query-source address options properly for each
>view (as per the FAQ). Unfortunately, I was never able to get the two
>servers to properly notify eachother, unless I restarted named on the
>slave server - then the transfers would work perfectly. So I created a
>horrible workaround of restarting the slave server's named process
>every 10 minutes (under the gun with a screaming boss).
>
>The follow-up message I received from Mark Andrews on 11/02/2003
>suggested that I use "rndc refresh" and check my notify-source. I did
>this, but ran into a snag with the "rndc refresh" command. Every time
>I ran it, I would get:
>
>rndc: 'refresh' failed: unexpected end of input
>
>I troubleshot the problem and eventually gave up - since my workaround
>seemed to be sufficient.
>
>I now have two completely different, fresh RedHat 7.3 +
>bind-9.2.1-1.7x.2 servers that I am using to work this out once and
>for all. I manually re-created the named.conf files and every zone by
>hand. However, I have only been able to get the system to the *exact*
>same status as the current live DNS servers - zones only update when
>the slave server's named is restarted. I also still receive the rndc
>refresh EOI error.
>
>So, I completely removed bind, config file(s), and the zones from the
>slave server. I reinstalled bind with a config file like so:
>
>options {
>        directory "/etc";
>        pid-file "/var/run/named/named.pid";
>        };
>
>zone "." {
>        type hint;
>        file "/etc/db.cache";
>        };
>
>I still receive the rncd refresh EOI error. So, I've done searches on
>various mailing lists and can't find a ton of information about this
>specific error. My system logs don't bark anything, and turning up the
>log mode in bind doesn't show anything either.
>
>The zone transfer problems aside, does anyone have any idea what's
>going on?
>
Well, you need to specify a zone for the "refresh" command, see "rndc 
-h". What did you think a non-zone-specific "refresh" would do? Refresh 
*all* zones? That would have been hardly better than your "horrible 
workaround"...

                                                                         
                                          -Kevin