>On Mon, Jul 31, 2000 at 02:13:44PM -0400, Ken Hornstein wrote:
>> A semi-related datapoint: adding the NTP kernel option makes the Shark
>> randomly hang (seems to be dependent on how much disk activity you do).
>
>Hey, could this be related to time going backward and callout() ?

It seems unlikely, since I wasn't running ntp on those machines :-)

Just the presence of the the NTP code in the kernel did it.  My
_theory_ is that since the stuff that now happens during clock
interrupts takes a lot longer, some critical interrupt is masked for
too long.  That's just a wild-assed guess.

Erik, what version of NetBSD are you running on your Sharks?

--Ken