odd bind problem

[Tim]

On Fri, 2009-03-27 at 14:16 -0400, Mark Haney wrote:
> I've got a newly installed F10 box that is running BIND as a slave.
> I've been seeing something really odd, though.  About every 6 or 7
> days DNS recursion fails.  There are no real error messages, but
> suddenly I can query any record we are SOA for (we host a couple dozen
> zones for customers) but we can't query, say www.google.com from that
> server.  We get a SERVFAIL.

Are you using NetworkManager?  (A really bad idea for a server.)  And
has an interface gone off-line and back on again?

There's a common problem with an awful lot of services that if their
network interface goes down, then up again, or just isn't up before they
first start up, that they don't use that interface.  The workaround is
to restart such services whenever NetworkManager changes state with a
script in the /etc/NetworkManager/dispatcher.d/ directory.

The issue of services breaking after network connections going up and
down isn't exclusive to NetworkManager, but it seems to make the
situation worse.

At the moment, I've got scripts in there for NTP and sendmail, since
those services on this laptop are affected by that problem.

-- 
[tim at localhost ~]$ uname -r
2.6.27.19-78.2.30.fc9.i686

Don't send private replies to my address, the mailbox is ignored.  I
read messages from the public lists.