About once a day our netbsd network server reboots right after a device
timeout. What I'm looking for is either a way to make the machine
respond in a less drastic way (just reinitialising the network card with
ifconfig down, ditto up usually helped in similar cases) or to prevent
the device timeout alltogether.

The kernel is a NetBSD - release, last updated on 11/25/99. (Tried
-current too last week, no relief).
The card, I know, none too bright, is a RealTek 8139 type clone. I fixed
it's settings at 10MB UTP, IRQ15 using the card configuration utility
since autodetect didn't seem to work. IRQ 15 should be okay as long as
2ndary IDE is disabled in CMOS setup and has no kernel support either.
I've also tried IRQ 9 but that is not exactly without risk either and
the card displayed the same problem.

A similar problem kept occuring on the old server running RedHat Linux
6.0 and using a Compex RL2000 interface card on IO 0x280 IRQ 5 (which
was enough reason to try and switch to NetBSD instead).

I would appreciate any suggestions,

Thanks, greetz and whatnot,

Martijn