Subject: Re: update(2): Re: netboot on 5000/240 probs
To: None <port-pmax@netbsd.org>
From: Toru Nishimura <nisimura@itc.aist-nara.ac.jp>
List: port-pmax
Date: 10/27/1999 11:18:51
>> There are hundreds of lines of:
>> ?IO: LANCE  ste 0xa6b3 - - - - - - - - - - - - - - - - - - - - - - - 
>> - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>      [ deleted ]
>
> LANCE status register complains there were network transaction error
> which should not have happen in correctly configured Ethernet network.

There is a possiblity of 'SQE switch is mistakenly ON' syndrome. The
SQE must be OFF for 'siginal repeating' devices like the 'repeater',
'bridge' or 'hub'.  (I'm afraid that many don't have knowledge about
what they are in this UTP damaged Ethernet world) If SQE is set wrong,
all transactions will cause jam condition on every transmition.  If
Network General's Sniffer is attached to such a misconfigured network,
it will show astonishing high rate of collision condition.

Standard diagnostic approach is to incrementally split the doomed network
into pieces until no error happens, or to build an absolutely minial
network which never brings any error then compose it with the rest of world
incrementally until any error happens.

Tohru Nishimura