BIND 4.9.11 problems on Solaris 8

[Gavin Hurlbut]

Hello, all.

I am in the process of upgrading a set of old BIND 4.9.11 servers to more
current hardware, then more current software (BIND 9.2.x).  As a starting step,
I have compiled and setup BIND 4.9.11-REL on a Solaris 8 machine (SPARC in
case it makes a difference) so we can do a "forklift upgrade" on the hardware
while planning for a more orderly upgrade to 9.2.x.

I was wondering if anyone out there had successfully run BIND 4.9.x (especially
.11) on a Solaris 8 machine?  The way it is now, if I start BIND using a script
in /etc/rc2.d, it seems to want to go crazy and fork many children.  Of course,
this causes BIND to run through all of the virtual memory, and fail, taking
with it anything that started after it (such as ntpd) as there is no memory to
be had.

I tried increasing the swap space from 2G to 4G, no difference.  I could get it
to reliably start up if I launched it from an "at" job, but this isn't really
sufficient.  It will also start successfully from a command prompt.

I'm hoping there's some system tuning or some tuning of BIND that was missed,
but there's no specific mention that I could find in the BIND-4 documentation
about Solaris 8, so I don't know what it would be.

Any ideas?

Thanks.
Gavin