>Not all of them have a reliable internet connectiom so using ntp
>isn't a great solution. It's probably a firmware problem rather than
>NetBSD per se (it would be interesting to find out if the ARMLinux
>people have seen it, I've CC:d Phil Blundell), but it would be nice
>to get the problem sorted.

I've never noticed it, but equally I never paid much attention to the clock.  

Are you talking about the Dallas battery-backed RTC or the internal softclock 
that the kernel keeps?  (That is, do you mean that the time is OK until you 
reboot at which point it's way off, or that while the machine is running its 
clock gradually drifts?)

I don't actually have a CATS at the moment so I can't look into it directly.  

p.