Tested on 5.0_RC2 running on a 4000/500A...
There was a thread a few weeks ago about system hangs and how breaking
into ddb and single stepping for a while would un-hang them. I've now
found a simple way to reproduce the symptoms: just run a program with an
infinite loop in it, other terminal sessions will stop responding but
the box will still respond to pings. This looks like the scheduler is
being locked out somehow but I'm unfamiliar with the kernel internals so
I'm hoping someone more knowledgeable might have some suggestions.