[H-GEN] Ethernet troubles

[Robert Brockway]

[ Humbug *General* list - semi-serious discussions about Humbug and ]
[ Unix-related topics.  Please observe the list's charter.          ]

Hi all.  I recently acquired (for free) 2 Tektronix xp10 Xterminals.  I
set them up on my home coax network (Its always worked ok up until now...)

Anyway I gleefully set them up and got the Xserver downloading using tftp.
I've done all this numerous times before and never had the problem
below...

At between 40% and 80% the tftp d/l stops and eventually timesout.

I did the usual and trimmed stuff off the net so that only the Xterm and
the boot server remained, thus:

 Xterminal-----Blake (boot server)

I then swapped the Xterminal,terminators,T-pieces,cable and even the boot
server, effectively ruling out any one point of failure (no one piece of
equipment is common to the problem).  In all of these cases the problem
persisted.  There was no appreciable difference except that using zen as
the boot server seemed to make the problem worse - no network activity in
some cases.

Everything appears properly shielded and in any case there is no common
element.

Anyway, I ran tcpdump and got the following results...

22:35:11.849107 blake.ralnet.net.4289 > avon.ralnet.net.1031: udp 516
22:35:11.851877 arp who-has avon.ralnet.net tell blake.ralnet.net
22:35:11.854197 arp reply avon.ralnet.net (0:0:b4:5e:ad:62) is-at
0:0:b4:5e:ad:62 (0:0:0:0:0:0)
22:35:11.857266 avon.ralnet.net.1031 > blake.ralnet.net.4289: udp 4

At this point the tftp d/l halts and never moves again.  Then after about
30-60 seconds, we get this:

22:35:49.461787 avon.ralnet.net.1031 > blake.ralnet.net.4289: udp 4
22:35:49.462152 arp who-has avon.ralnet.net tell blake.ralnet.net
22:35:49.464087 arp-#1031 for proto #2048 (193) hardware #1 (133)
22:35:50.461918 arp who-has avon.ralnet.net tell blake.ralnet.net
22:35:50.463861 arp-#0 for proto #2048 (125) hardware #1 (26)

This behaviour is highly reproducable.

A couple of other things to note.  I did actually get it working once but
repeating a boot numerous times with the same configuration failed to
repeat the success.

Also, putting zen and blake (both Linux boxen) on the lan alone results is
0% packet loss on a ping flood from blake to zen, indicating that the lan
is ok in that state (apparently).

I hope this is clear as I sam quite confused :)

The only conclusion I can draw right now is that the Xterms are broken by
design as they both exhibit the same behaviour but that seems unlikely.

TIA for any help, this one has me stuffed.
Rob

--Robert Brockway B.Sc.  Email: robert at ralnet.net, robert at humbug.org.au
                                robert.brockway at uq.net.au
			 WWW:   http://ralnet.net/~robert
Founder and current Librarian of HUMBUG (http://www.humbug.org.au)


--
* This is list (humbug) general handled by majordomo at lists.humbug.org.au .
* Postings to this list are only accepted from subscribed addresses of
* lists 'general' or 'general-post'.