forwarder and cache

[/dev/rob0]

On Mon, Mar 16, 2015 at 10:36:40AM -0700, Dana Huggard wrote:
> I have a problem with a host lookup failing, but then succeeds 
> after I restart bind9.  The reverse look succeeds before the 
> restart.
> 
> There are two bind servers.  A and B.
> Server A is master for A.domain and B is master for B.domain.
> Server A has a ZONE B configuration entry declaring B as the
> forwarder.

It's usually better to use regular DNS delegation to make this work 
properly.

> Server B also runs dhcpd with ddns.  A new computer comes up via 
> dhcp on the B domain and network.  If I query the hostname using
> B as my resolver I get and answer, If I query the hostname using
> A as my resolver I get NXDOMAIN.  If I then restart bind on A,
> and re-run the same query that failed before, it now succeeds.

Restarting is overkill.  Try "rndc flush".

> I don't understand why this is behaving this way.  Any ideas?

My best guess here is negative caching?  If you (or any user of 
resolver "A") had queried that name in zone B during the period 
defined in zone B's SOA "minimum" field (the last numeric field in 
the SOA), the NXDOMAIN result is cached.

For more help show your actual dig commands and results.
-- 
  http://rob0.nodns4.us/
  Offlist GMX mail is seen only if "/dev/rob0" is in the Subject: