This is probably a local bug, but it seems to work on every Sun on our
network (and our network is relatively "tight" -- i.e., the sysadmins have
done just about everything in their power to tighten the security screws
down on the system):

Each Sun has /dev/audio as a connection point to the speaker.  This is used
by play and a few other sound related commands.  If you're lazy you can just
cat to it or cp to it to dump sounds from files (I was using this to annoy
the hell out of someone when I found the upcoming 'reboot' trick).  Anyway,
doing this locally (i.e., while logged directly into the machine you're
dumping sound through) causes no problems.  The obvious extension to this is
to us% something like 'rsh' to dump sounds on every machine on the network
(that way everyone gets a rousing rendition of "Paint it Black" at 3 in the
morning....).  So, I type:

rsh someothermachineonmynetwork "cp ~/sounds/PaintItBlack /dev/audio"

and the command "hangs".  Not wanting to walk down the hall I just
rlogin to another machine an$ do the same rsh to the workstation in my
office... I hear a bit of PaintItBlack, then the machine panics, dumps core,
and reboots.  Don't know for certain why (don't know whether this is a
global phenom, but it works here).

Anyway, now when some schlep is running 18 processes, has filled /tmp to the
max, and taken a vacation in Tahiti (and I don't feel like driving 15
minutes to the office to flip a damned switch) I just login on the network
and say   rsh machine "cp ~/.cshrc /dev/audio", wait about 10 minutes and
log into the freshly rebooted machine.

Give it a try -- I'd be interested to see whether this works on other nets
or not....