Bind 4.9.11 under Solaris 8

[Gavin Hurlbut]

Hello all.

I'm in the process of performing a series of upgrades to take some BIND 4.9.11
servers and upgrade them to BIND 9.2.x.  The first stage is to do a "forklift
upgrade" by replacing the ix86 Solaris 2.6 boxes with Sparc Solaris 8 boxes,
while still running BIND 4.9.11.  This is being done to maintain the current
service level while planning a smooth transition to BIND 9.2.x.

I have compiled it, and have it starting up from a script in /etc/rc2.d,
however it seems to have troubles starting.  The system runs out of swap
resources, even with 4G of swap.  It seems to spawn several hundred forked
children, causing named to run out of memory, along with the rest of the 
system.  However, if I run it from a shell, or start it from an at job, it 
starts up just fine.  Once it is started, it seems to be operational.

Has anyone out there got any idea what might be causing this behaviour?  It 
seems to happen during the initial zone loading, however, I don't understand
why as I don't see any loops with fork() or variants in it in my quick scan
through the code.

While I realize that not many people are still running 4.9.x, I'd really like
to get these servers running so I can retire the old machines that are getting
rather aged.

Any ideas?

Thanks
Gavin