a/ux 1.0.1 (actually, ethernet interfaces)

[Tim Tessin]

In article <7325 at bloom-beacon.MIT.EDU> dyer at arktouros.MIT.EDU (Steve Dyer) writes:
> 
> I would guess that you don't know enough about the situation to say that
> there is no bad Ethernet card situation.  The symptoms seem to relate
> to more than one packet arriving at the interface within a short period of
> time--it's not a protocol problem, though it may be a device driver problem
> or limitation.  I wouldn't dismiss hardware, however.  I've seen this happen
> under A/UX with (presumptive) broadcasts but also with a simple
> "rcp -r remote:dir dir"--the network dies every time.
> 
> That is not the case here in any of the recent postings.  I would like to
> know, of those of you who are running A/UX on a heterogeneous ethernet,
> if there are any of you who are operating WITHOUT this problem.  That
> would be a useful datapoint.

I bought a Kinetics EtherPort II card for my Mac II.  The kinetics card
is not a 3com, it's an Excelan somthing-or-other.  I called up
Kinetics and they sent me a A/UX driver.  It hung just like the Apple
cards.  A few calls to tech support and they had the guy who wrote
the driver out here to fix it first hand (Of course, they are only 30
miles away).  The problem seems to stem from other machines on the
network not obeying the minimum packet inter-arrival specs for
Ethernet.  I suppose you could say that machines who don't obey the
spec don't belong on the net, but that ain't realistic.  We have a very
large network with about 1500 machines from pc's to cray-2's. The big
machines provide a backbone using hyperchannel with DEC LANBridge 100
to bridge to the ethernet sub-nets.  Since we run heterogeneous
protocols (DECNET, TCP/IP, etc) over the wires, we can't filter the
packets, so everyone sees everything on the net.  My A/UX system hasn't
quit since the driver fix, running uninteruppted for at least 4-6 weeks
at a stretch (that being the longest I have run it without voluntarily
rebooting).

Tim Tessin - Lawrence Livermore National Laboratory  
PHONE: (415) 423-4560 
ARPA:  tjt at tis.llnl.gov
UUCP:  {ames,ihnp4,lll-crg,lll-lcc,mordor}!lll-tis!tjt