[Mimedefang] mimedefang process dies when system runs out of swap

[Scott.Larnach at ed.ac.uk]

Solaris 5.8, sendmail 8.12.6, mimedefang 2.20

I've been load testing my system and have discovered that the mimedefang
process dies when the system runs out of swap. (I have put my order in
for more memory.) Is this a known problem?

Of course it would be desirable for it to continue running so that the
system would recover after such a condition rather than having to
restart mimedefang by hand. So I thought I'd have mimedefang dump a core
and see what the stack was like.

I changed the chdir("/") in mimedefang.c to chdir("/tmp") so it can
write the core (and fuser tells me that mimedefang does have /tmp open
as cwd). Took the strip out of the Makefile. Remade mimedefang.  Added
"ulimit -c unlimited" at the top of /etc/init.d/mimedefang. But no core
dump even when I kill it with an ABRT signal. I think I must be missing
something and if anybody can see what I'd be grateful. Thanks,

Scott Larnach, Edinburgh University Computing Services