sh: Cannot vfork

[Simon J Gerraty <sjg%juniper.net@localhost>]

I get a lot of email...

I've been having increasing problems with mail delivery due to:

Dec 17 11:41:07 chaos postfix/local[28962]: 0396A580A9:
to=<sjg%chaos.jnpr.net@localhost>, relay=local, delay=396, delays=396/0.02/0/0.01,
dsn=5.3.0, status=bounced (Command died with status 2:
"/usr/pkg/bin/procmail || exit 75 #sjg". Command output: sh: Cannot
vfork )

I'm not sort of ram,swap, or disk
even when I see fork errors on console, top shows <= 200 processes running
similar to the output below.

Last night I was unable to login to this box - ssh failed to get a
shell.

I have increased the max number of processes that can run both globally
(3k) and per user (1k)

I have tweaked /etc/postfix/master.cf to limit the number of instances
of local (21) and smtp (42)

I even upgraded the OS from  5 stable (otherwise very solid) to
NetBSD chaos 6.1.5 NetBSD 6.1.5 (GENERIC) amd64
but the problem continues.

$ ulimit -a
time(cpu-seconds)    unlimited
file(blocks)         unlimited
coredump(blocks)     unlimited
data(kbytes)         3145728
stack(kbytes)        65536
lockedmem(kbytes)    1305017
memory(kbytes)       3915052
nofiles(descriptors) 128
processes            1044
vmemory(kbytes)      unlimited
sbsize(bytes)        unlimited

$ pstat -hT
1062/3404 files
 202531 vnodes
 432M/2.0G swap space

$ top -u | head -8
load averages:  0.02,  0.04,  0.08;               up 94+01:26:50 08:23:35
195 processes: 192 sleeping, 1 stopped, 2 on CPU
CPU0 states:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt, 100% idle
CPU1 states:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt, 100% idle
Memory: 1048M Act, 19M Inact, 22M Wired, 82M Exec, 528M File, 1494M Free
Swap: 2048M Total, 432M Used, 1616M Free

  PID     UID  PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU   COMMAND
  
Thoughts?