100% cpu on linux bind 9.3.0 AND 9.2.4

[Robert Weber]

I am running a bind server on a new linux server.  It is a EMT P4 box from
Dell running Redhat Enterprise 3.0 64bit.  I complied with -O3 and -O2,
with and without threads(but it didn't seem to matter) and both current
bind9 trees.  I was running on a Ultra10 with solaris 8, and even though it
was 1/6th the machine it was around 60% cpu.  On the linux box I get 100%
usage and I can't figure it out.  I don't have signed zones/queries but I
do have some dynamic zones and 2 views.  There isn't much in the logs other
than the usual denied updates from clients.  Tcpdump doesn't show much in
the way of traffic either.  rndc status doesn't indicate any large number
of recursive queries or anything and I am stumped.  Any ideas?  

--
Robert Weber, UnixOps/ITS UCB

"It was like the Manifest Destiny all over again, except, instead of taking
and consuming everything in their paths for God, they did so with the same
fervor and sense of entitlement for their new god... themselves. Buy, take,
break, throw it away."  --FIF "American Kryptonite"