two primaries and no slaves - good or bad?

Barry Margolin barmar at alum.mit.edu
Wed Jul 28 22:54:05 UTC 2004


In article <ce983o$et2$1 at sf1.isc.org>,
 Hugh Beaumont <hbeaumont at yahoo.com> wrote:

> Due to various issues, I've considered doing this :
> 
> ns1 is a master/primary
> 
> ns2 is a master/primary (I think I have my terminology a bit messed up)
> 
> ns2 rsyncs /var/named/ and /etc/named.conf every so often from ns1 and then 
> reloads itself.
> 
> 
> 
> Is there anything wrong with doing this?

Nope.  As long as all the servers sync up periodically, no one cares how 
you arrange it.  Zone transfer is just the mechanism that happens to be 
built into the DNS protocol, but if you can achieve the same results 
some other way, it's fine.

> Is it true to say that a slave really only exists to allow automated updates 
> to a nameserver you
> don't necassarily have full acceess to - for example a peer nameserver who 
> agrees to be your slave
> and can get updates from your master nameserver through zone transfer, etc.

Correct.

> Is it true that a client has no idea if the server they are talking to is a 
> master or slave? 

Yes.  The distinction is only important to the DNS administrators.

-- 
Barry Margolin, barmar at alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***


More information about the bind-users mailing list