Whacky Lite-On craziness

Corbett J. Klempay cklempay@acm.jhu.edu
Wed Dec 2 15:42:05 1998


Hey all...I'm new to the list.  My ACM chapter recently bought a bunch of
machines to assemble a cluster with (8 PII-350's).  We're trying to bring
the master online, but it is complaining like a bitch...it has 2 Lite-On
cards in it.  We have to only have one card in it to get an ftp install to
go ok...that works fine...but once we plop the second card in, we lose all
network access.  We edited the /etc/sysconfig/network to add eth1 and
all...and if you try to shut down networking and then bring it back up (to
get eth1 to come alive), the whole machine dies and all you see is a
"Divide Error: 0000"...and this only happened in our original config (1
Lite-On, 1 Tulip in the same machine)...when we switched to 2 Lite-Ons
(by swapping with one of the compute nodes), it doesn't die, but instead
we still cannot get outside access.  One guy found that if he rebuilt the
kernel without modules and explicitly added a route to one of our other
machines, he could get to it.  When he tried building with modules, he
couldn't get to the other machine at all (even with the explicit route).
We are using version 0.90 of the driver (and had previously tried whatever
it was that came with RH 5.2...I think .89H).  We see messages such as:

"
Corbett:  here are the error messages:

eth0:  The Transmitter stopped!  CSR5 is 2678016, CSR6 is 816e2002
eth1:  The Transmitter stopped!  CSR5 is 2678016, CSR6 is 812e2202

and it displays these right at the login prompt ~2 seconds after the login
prompt is displayed.

I get it with either kernel.  It gives different numbers sometimes
though:  I've seen

	CSR5	2068012
		2678016

and 

	CSR6	816e0002
		812e2202"

If it matters (someone suggested it might), we have a Promise USCSI card
(NCR 875-based) in the machine.

------------------------------------------------------------------------------
Corbett J. Klempay			         Quote of the Week:
http://www2.acm.jhu.edu/~cklempay  "Advice is what we ask for when we
				    already know the answer but wish we
				    didn't." 

PGP Fingerprint: 7DA2 DB6E 7F5E 8973 A8E7  347B 2429 7728 76C2 BEA1
------------------------------------------------------------------------------