"most field lockups i've seen have been caused by an overheated CPU, which
is fixed by a reboot."
After reading this I did some checking and found very high temperatures in
my units - all of them 1 rack unit, mounted on top of each other in a
cabinet that houses a lot of other networking gear (the cabinet is
pressurized for cooling). I found the most used box to be very hot to the
touch (mostly around the RAM) - hot enough that while I was spacing them
apart it hurt my hand while holding it. So, after putting one rack space
between them all (3 m0n0wall's in this cabinet), they have been running a
Just over 5 days now and no issues. I will update as the days go on.
Thanks Dinesh for the idea.