Hm, this actually gave some ideas about what may be happening:

> > 
> > ----Goblin.BQTnet.SE PING Statistics----
> > 14375 packets transmitted, 14375 packets received, 0.0% packet loss
> > round-trip min/avg/max/stddev = 3.950/4.846/197.504/8.889 ms
> > 
> ----Goblin.BQTnet.SE PING Statistics----
> 11039 packets transmitted, 11012 packets received, 0.2% packet loss
> round-trip min/avg/max/stddev = 3.842/5.703/4812.884/67.312 ms
> 
Look at the VMS numbers: the max time is 197 ms, which probably indicates
that the aliveness of the card is checked after 200ms, and when the
average is a little bit higher this restart have happened a couple of times.

The NetBSD numbers show the same effect, but the restart happens after
5000 ms. This is much longer time, so come constructive use of dc tells
that there were a reset 4 times during this period.

Apparently VMS does its alive check closer to the transmission start, 
therefore no problem is normally noticed. Maybe someone with the VMS
sources close by can check this? :-)

-- Ragge