Subject: RE: wiring computers with 10BT -- progress
To: April G Gruhn <gudongfe@pilot.msu.edu>
From: Laine Stump <lainestump@rcn.com>
List: netbsd-help
Date: 10/03/1999 18:47:52
At 05:17 PM 10/3/99 -0400, April G Gruhn wrote:
>I have got the two ethers installed and I configured them by adding
>lines to /etc/netstart.local in both the i386 and the i486.

The officially sanctioned way to do this is to add a file called
"ifconfig.<devname>" in /etc containing all the parameters on the ifconfig
line past "ifconfig <devname>". So, for a device called ep0, you'd add a
file called /etc/ifconfig.ep0 with the following info (for example):

	10.0.0.1 netmask 255.255.255.0

>Here is the $10000 question: How do I get them talking to each other?
>I assume I will need two routes set up to do  the routing:
>A --->B
>B --->A  correct?

If they're on the same subnet, you won't need any extra routes. For
example, if the ethernet on both machines is ep0, on machine A, just give
the command:
	
	ifconfig ep0 10.0.0.1 netmask 255.255.255.0

and on machine B the command:

	ifconfig ep0 10.0.0.2 netmask 255.255.255.0

(or create /etc/ifconfig.ep0 on each machine).

Since they're on the same IP subnet, they'll find each other with ARP (and
will also find any other machine between 10.0.0.1 and 10.0.0.254 with ARP).  

You'll only need to add a route if you're trying to contact a machine on
another subnet.

If you want to verify that there are packets leaving each machine, and that
they're "seen" at the other end of the cable, run "tcpdump -n -i ep0" on
both machines. When you attempt to ping one machine from the other, you'll
first see an ARP request, then an ARP response, then the icmp (ping)
packets will start. If the ARP request sent by one side isn't seen by
tcpdump on the other side, you most likely have a cabling problem (although
it could be something like an improperly jumpered card).

>The goal here is to telnet into the i386 and access the SCSI-2 devices...

Start with ping. (As a matter of fact, use "ping -n <address>" so that you
don't mistake broken DNS services for a broken network).

When you get to telnet, the machine acting as the server will either have
to have DNS disabled (by having no resolv.conf file), or you'll end up
waiting for 30 seconds or so before the login prompt comes up (while
telnetd is attempting to reverse resolve the client's IP address to a
domain name).

If the SCSI device is a disk, then you'll need to setup one machine to be
an NFS server and the other to mount its disk. But first get the machines
talking.

BTW, I've redirected this message from port-i386 to netbsd-help@netbsd.org,
which is the proper venue for questions of a general nature.