Subject: more data Re: 1 of 2 Ethernets suddenly stopped working?
To: None <,,>
From: Der Beagle <sopwith!>
List: port-alpha
Date: 11/22/2001 15:50:07
More data on my 1 of 2 Ethernets suddenly stopped working problem:

	Having the host send out a ping to either of the
	broadcast addresses appears to work fine.  The
	host even responds to itself.  What is special about
	the broadcast addreses?

	Having the host send out a ping to addresses other than
	the 2 broadcast addresses or itself do not (usually) increment
	the packet counts as reported by netstat -i.  If the
	packet count does increase, the output error count
	increases as well.  (with de driver)

	Having an X terminal send a ping using the host's IP
	address generates an error message on the host complaining
	about the duplicate address.  Therefore: the network hardware
	is working, and the packets are making it into the kernel,
	and the kernel is doing something with at least some of them.

	There were no software or hardware changes between working
	and not working.

It is not clear to me why the kernel would suddenly stop responding
to tftp requests on one network but still work fine on the other.

It is not clear to me why the kernel would not respond to pings,
except to complain if the address were a duplicate.

It is not clear to me why outgoing packets do not always show up
on netstat -i, even though LEDs on the transeiver flashes,
indicating that the packet went out.

Hubert writes:

> AUI? try T-piece (or what's that called, that you attach tp tje
> transceiver).

I haven't tried swapping out the BNC tees, but the hardware is working (see above).

Bernard writes:

> The most common cause of this kind of problem I ever saw back when
> I had to look after 10base-2 networks was that someone had pulled
> one of the terminators


> As I remember it, resistance (what you probably measured with your
> ohmeter) isn't a fantastically useful measure of the health of your
> network (unles you're seeing open or short circuit). I think the
> terminators are actually providing impedance (but I'm not an electrical
> engineer, so I may well be wrong on this).

Yes, I was measuring DC resistance.  I am an electrical engineer, and yes
there is more to it than that.  I was just checking for opens and shorts.
Anyway, by using a dup IP address, I found that packets do make
it across the network.

The bad network has the host, an AUI-to-fiber-to-AUI link, and a 10base-2
network with 2 X terminals.  Very simple.