New Serial # Is Less Than Old Error

[Burkhard Weeber]

In addition to James notes there may occur a problem if you try to refresh
your named process by issuing a "kill -1 <PID>" altough this is legal in the
manuals.

On my AIX machine I have discovered that if I "kill -HUP" the process is
immediately terminates without dumping the database to the zone files. If
you are doing DDNS your secondaries will then be ahead of the primary
thowing this error.

Terminating the named with "kill -TERM(15)" is OK, the zone files are
completely dumped before the process terminates.

HiH

Burkhard Weeber
viastore systems GmbH
P/O Box 300668
D-70446 Stuttgart
Email: B.Weeber at viastore.de


-----Original Message-----
From: bind-users-bounce at isc.org [mailto:bind-users-bounce at isc.org]On
Behalf Of James Raftery
Sent: Thursday, January 11, 2001 3:42 PM
To: Bind Discussion List (E-mail)
Subject: Re: New Serial # Is Less Than Old Error



On Thu, Jan 11, 2001 at 09:27:58AM -0500, Smith, William E. wrote:
> I've begun seeing in my messages log the following error:
> new serial number < old (88954 < 88957)

The message should also mention the name of the zone with the problem.
When you reloaded that zone BIND noticed that the serial number on the
newly loaded copy of the zone was smaller than the previous one. This
shouldn't happen. Serial numbers should be increased and (almost) never
decreased. It is by increasing the serial number that you allow your
secondaries to tell if the zone has changed since they last checked.

The serial number is one of the fields of the zone's SOA record. To
solve the problem, query for the SOA record of the zone from each of the
secondary servers and note the largest serial number you see. Edit the
zone file on the master and set the serial number to be one increment
larger. Then reload the zone.

james
--
James Raftery (JBR54)
  "Managing 4000 customer domains with BIND has been a lot like
   herding cats." - Mike Batchelor, on dns at list.cr.yp.to.