DNS queries go to primary and secondary DNS servers at the same time

Chuck Aurora ca at nodns4.us
Tue Dec 17 17:42:01 UTC 2019


On 2019-12-16 13:13, Roberto Carna wrote:
> I have a primary and a secondary BIND9 DNS servers, working as master
> / slave with zone transfers between them.

Primary/master and secondary/slave are concepts which apply only to
authoritative servers, and in this case you are talking about these
servers acting as resolvers, not authoritative.

> I have several Linux machines (desktops and servers) with Debian and
> Mint.
> 
> I've realized, using TCPDUMP at DNS1 and DNS2, that all DNS queries
> from Linux machines go to both DNS1 and DNS2 BIND servers at the same
> time.
> 
> In all Linux machines we have setup the DNS resolution using
> /etc/resolv.conf:

And resolv.conf is not an ISC BIND thing.  In GNU/Linux it's usually
GNU glibc which uses that file.

> nameserver IP_dns1
> nameserver IP_dns2
> 
> But when the from Linux clients I execute:
> 
> $ host <some_fqdn>
> 
> I can see UDP traffic arriving to DNS1 and DNS2 at the same time.
> 
> What can be the problem ?

I'm not sure I would see that as a problem, although you might want to
simplify and use only a single resolver IP address.  If your site is
big enough to need two or more resolvers, use an anycast address.  For
example, Google's 8.8.8.8 is a large farm of nameservers distributed
throughout the world.

> Because I expect only DNS traffic going to
> DNS1 because it is before DNS2 in /etc/resolv.conf.

GNU glibc does have documentation, starting with the resolv.conf(5)
manual.  I'm not sure if there is a specific mailing list or forum to
discuss it, however.


More information about the bind-users mailing list