Tom writes:

> I am convinced that I have some kind of intermittent hardware problem. For
>  no apparent reason, the 918 is talking to the LAN again. I had reseated the
>  SCSI/console/lan card, restarted the 918, and still could not ping or
>  connect to it. A few minutes later, a spool file started printing on the
>  networked printer, I can ping the 918, and the 918 can ping other devices
>  on the LAN, even though I had done nothing after restarting the system.
>  It's got to be some hardware problem somewhere.

The trick in these sorts of situations is to create an absolutely minimal LAN
with the HP3000. That minimal LAN might be a very short segment of ThinLAN
coax running from the bulkhead of the back of the 918 to (i) a DTC (and no
further) or to (ii) a BNC on the back of a hub (and no further). If (i) a
regular terminal can talk to the HP3000 reliably through the DTC or (ii) a VT
or telnet-based terminal emulator can reliably communicate with the HP3000,
you know that the HP3000, its NIC card and the short coax cable are all OK.

But if they can't, you also know that the problem is in one of those
components. Nonetheless, you now have a system simple enough so that you can
change out the various components until you find the flaky one. If you built
your own coax cable, suspect that first. They tend to be intermittently
unreliable. But if you purchased the cable from a major supplier, they rarely
go bad.

The trick in any debugging process like this is to first greatly simplify the
problem, then surround it and kill it (oops, sorry that was Colin Powell's
tactics in the last Gulf War. Too much CNN-generated testosterone, I guess).
The real trick is to first greatly simplify the problem, using an absolute
minimum number of components, and then systematically generate hypotheses
regarding which must be the failing component, ordered principally by the
known unreliabilities of the components, switching them out with spares,
until you have a working system again.

I just very recently installed a LAN at one of our long-time customers here
in town, although we had all of the inter-office, in-the-wall wiring done by
a professional contractor. Approx 3 weeks after the LAN was up and working,
it too began failing in an intermittent manner. The cause was the BNC
connectors on the ends of the long coax cable that ran from the DTC to the
24-port hub. The installers used a crimping tool instead of soldering the
center pin onto the center conductor. Once a new cable was restrung and
properly put together, it's been as solid as a rock since.

Wirt Atmar

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *