This maddening problem continues...on this 5000/25

Attempts to boot into multi-user almost always fail.  This is what
happens...everything proceeds normally until (most often-sometimes
it fails during the building databases, which comes right after the
checking quotas):
....
checking quotas: done.
Date/time maxmin init: /bin/sh on /etc/rc terminated abnormally, going
into single user mode

This usually results in a sh.core, but sometimes I also see core files
for portmap and syslogd.

There are no other messages.  After responding to the terminal type, 
attempting almost anything results in a trap and into the debugger...
so, I do a 'reboot'...another attempt to boot to multi-user at that
point results in the same thing...the only thing that seems to break this
deadly embrace is to power it off, then back on, and boot to single user.

if I boot to single user, then fsck the filesystems, they usually
are reported as clean, then a Ctrl-D and on to the rest of the boot and it
succeeds more often than not, and from that point on everything is fine.

I thought for a time that it was a cold start thing, so I've let it sit
for up to an hour after powering it up, and it still happens.  Over the 
past two weeks, there has been only one successful boot to multi-user from
a power up, and that was after it had been up and running most of the day,
and had been shutdown for less than two hours..I did this as a test to see
what happened...

I don't recall booting into single user on that initial attempt after 
power
up over the past week, so I'll have to try that to see what happens.  In 
the past that didn't change things as it often resulted in the same sort
of failure after exiting single user...

Another thing, yesterday after getting it up I installed and ran the Byte
Benchmark test sequence...while not a hardware diagnostic, it certainly 
puts a good deal of stress on the running machine...1st, I ran it in an
xterm, and then again later on the console.  It ran thru the benchmark
without any problem, both times.

I would think that if there were a hardware bug that it would show up with
this, wouldn't it?  But, as has become the norm, once it is up, it stays
up...its just getting it there in the 1st place...

?
Paul