Re: Kernel OOPS with 1.2.11

Stephen C. Tweedie (sct@dcs.ed.ac.uk)
Tue, 1 Aug 95 13:18 BST


Chris Woods <cjwoods@gigotech.net> writes:

> On Wed, 26 Jul 1995, David C. Niemi wrote:
>> So it is quite possible to have long uptimes on NE2000s, at least with
>> newer kernels.

> And without major, significant network load. Well, maybe the NE2k drivers
> are stable in 1.3.x, but since all my machines are mission-critical (we
> lose machines, 4000+ paying customers don't get service) I cannot afford
> to run 1.3.x.

There's no doubt that early 1.2 kernels had big problems with their
ne2k code. However, 1.2.10 and later do seem to have cured the
troubles I was experiencing.

Earlier kernels could be brought to their knees by any sustained
traffic. The best test was to do a large rdist (we were trying to do
150-200MB rdists quite regularly at one point); the ne2k machines
would not survive more than about 10 minutes of this, usually. We
haven't seen a single ne2k hang since 1.2.10 (yet...).

Cheers,
Stephen.

--
Stephen Tweedie <sct@dcs.ed.ac.uk>
Department of Computer Science, Edinburgh University, Scotland.