A networking tech confirms: Miracles can - and will - happen!

Submitted by: Anonymous – Thu, 01/11/2007 – 17:53

Flashback to the early 1980’s. I’m a relatively new tech working out of a remote one person field office. My task, to install four brand new dumb (and I do mean dumb) terminals on an existing mid range system. The system is located in the business school of a university, and it typically has all 20 or so terminals occupied with students. The mini-mainframe has a processor that looks like a microwave oven and four washing machine sized drives in one room. The four dumb terminals will go in an adjoining room on the opposite side of a concrete wall.

I arrive onsite and unbox the four terminals. They all power up and pass their very basic POST sequences. The maintenance folks at the university have already pulled four runs of coax through a conduit to connect the terminals. All I have to do is terminate the coax, connect the terminals, and get them up to a login prompt. Heck, they’re so simple, there’s not even any type of network address or ID to set. Just turn them on, there’s a funky little growling noise, then a screen appears that says “Welcome to the ARC”. ARC being Attached Resource Computing. I connect the four units, power up the first one and get the familiar growlllll, Welcome to the ARC. Power up number two, growlllll Welcome to the ARC. Power up number three….and the entire system halts! Rats, I’ve got a DOA terminal and 20 irate students. I power everything down, go into the computer room and IPL (remember those?) the system. Ten minutes later everything is back up and running normally. I go back into the room with the four new terminals and power up the last one, growlllll Welcome to the ARC. OK, it works. Now I’ve got to troubleshoot the defective one.

These terminals are super simple, just a power supply, logic board, display board, keyboard and picture tube. Figuring it has to be the logic board, I order one for overnight delivery. The next morning I’m back at the university with logic board in hand. I put it in, connect the coax and power up. Growlllll, Welcome to the ARC. Success! I reach over to the terminal next to me, power it on and growlllll, Welcome to the ARC. I lean over to the next terminal in line; flip the power switch….and the entire system halts! AGAIN! Now I’m confused. This thing was working fine yesterday. Could the logic board have fried when I powered up the “defective” unit yesterday? That’s about all I can think of, and with the symptom being a complete system halt, I’m not about to spend much time experimenting. So, another logic board goes on order.

Day three, the logic board gets replaced in the last unit that caused a system shut down. I power it on and growlllll, Welcome to the ARC. So far so good, but I’m not celebrating yet. I cautiously power up another terminal. Ok, there’s the familiar growlllll, Welcome to the ARC. I let them sit for a while, thinking it might be some type of heat or load related problem. Everything stays up. Finally, I get my nerve up to hit the power switch on another unit, and….the entire system halts! Now I’m not confused, I’m mad. It has to be a problem in the coaxial hub all along! As soon as a third unit is powered on, regardless of order, everything crashes. There never was a bad terminal. OK, re-IPL, apologize profusely to the professor and the students for the lost work, downtime, etc. and get a replacement hub on order for same day delivery.

Hub arrives late that afternoon and, thankfully, the computer room is empty of students. I replace the hub and confidently march into the adjoining room to power up those four troublesome terminals. Number one, growlllll, Welcome to the ARC. Number two, growlllll, Welcome to the ARC. Number three, uh, number three? Hello? Total system halt!

All I can do is pick my jaw up off of the floor and try to put the oozing brain matter back into my head, as I think about updating my resume. What in the world could be causing this? Well, in desperation I begin to more or less randomly disconnect cables at the hub end and try different combinations of terminals. Finally I power on one terminal, then think to myself “Dummy, you’ve got this coax disconnected at the hub. This thing isn’t going to work”…. growlllll, Welcome to the ARC!!!!! I FLEW into the other room, thinking that someone was in there messing with the cables. But low and behold, the very same cable that I had just used on that terminal was lying on the floor, disconnected. I had just witnessed a miracle! A disconnected terminal had just communicated with its host system! And this was long before the days of wireless. A few more minutes of head scratching and I had an idea. I moved all four terminals into the same room as the main system, built some short coax cables, and connected them up. Glory be, they all four came up and worked like champs! I called maintenance and told them to strip out the brand new cabling they had run, and put in some new coax FROM A DIFFERENT SPOOL. Yes folks, believe it or not, I had fallen victim to what is undoubtedly the only spool of badly shielded coax to ever leave the factory.

Oh, and the mysterious terminal that connected while its cable lay disconnected on the floor? Simple RF leakage from one of the other cables while they were running through the same conduit together.

Comment:  Ghost in the machine

Everyone has a ghost in the machine story. The testing and troublshooting that goes on to figure it out are half the fun, unless its a prankster at work ... sometimes the ghost becomes a moving target

Jeremy (not verified) – Thu, 01/11/2007 – 22:14

Comment:  Ghosts

Yeah, mechanically defective things can be a royal pain to troubleshoot.

Back in the days of Arcnet and when I didn't know all that much about hardware: A stupid manager bought a machine at auction. Plug it in and down goes the network and I get called. Ok, it must be conflicting with the server. I pull the card, change the switches to something I don't think is in use, try it--works. I put it back on the guy's desk--and down goes the network. I try a different address, same behavior.

Ok, I'm stumped, it's off to the shop. Diagnosis: Address conflict, they changed it. Back on the guy's desk, down goes the network.

The final fix: The machine was a bit warped. When you tightened down the screw on the card something got torqued and apparently leaked voltage onto the cable, taking down the network. I reassembled it with the screw slightly loose and it ran fine.

Loren Pechtel – Thu, 01/25/2007 – 16:22

Comment:  ARC and network oddities

As I started through the article, I figured it might have been about a predecessor or a technology I worked with - ARCnet, which connected PCs and Servers - and every node had an 8-bit Node ID, set with a DIP switch. I took over a law firm network, which had problems with stations being blown off the network at seeming random times - the problem was duplicate node numbers. Fortunately the LAN was Novell, and I was able to use the info utility to get lists of node numbers and user IDs, and changed enough IDs so that each network branch had ID's which did not duplicate each other. I also learned to read the number a dip switch was set to at a glance!

Anonymous – Fri, 01/12/2007 – 12:06

Comment:  ArcNET

I used to work at a lawfirm using Arcnet. Beautiful stuff.

One of my successors started moving PCs from one segment to another and couldn't figure out why they would sometimes fail to connect to the network. Or maybe another machine would fail to connect. A friend told me about it, and I knew immediately why. But I wouldn't tell because she would have told the clueless tech.

Anonymous (not verified) – Fri, 01/12/2007 – 13:44

Comment:  ARCnet stuff

Nope, not ARCnet. Datapoint's proprietary ARC (Attached Resource Computing) LAN running their proprietary software. And as to whether or not it was RF leakage, I'm open to other suggestions as to how a totally disconnected cable managed to pick up a signal and boot a terminal! Keep in mind that after it displayed the Welcome screen, the terminal wouldn't perform any other tasks, but it was able to pull enough data to get the Welcome screen up.

Anonymous (not verified) – Tue, 01/16/2007 – 12:50

Comment:  similar story

Brings to mind a client of mine who was moving his computers from an office literally just off the truck garage to a new facility. When we cleaned the up to 2" of dust (I'm not kidding) out of the cases, one of the PCs stopped working. The RAM slot was totally 'bent' but the memory continued to work 'cause the dust was acting as a conductor.

VickyS – Thu, 01/25/2007 – 16:21

Comment:  I call BS on the RF leakage

I call BS on the RF leakage line. There simply would not be enough signal strength, not to mention proper impedance matching, to do this.

Anonymous (not verified) – Mon, 01/15/2007 – 11:24

Comment:  Call Your BS and raise it, too. Dn'tfold1stpostsincejan!

Let's just say that i know more about BS than i do RF. But still have encountered plenty of both. Live with, like the rest of us. But i gotta ask- when you call BS- is that before or after you start shoveling your own?
How do you know what the signal strenth and impedance were on those lines? Please do not try to explain to me how you would know if the impedance matched either, because you'd have more information? wouldn't you?
What i came away with from the initial post was that the technician possibly didn't tell us if or about any testing he did on the coax, and that he probably didn't have access to the test equipment available to techs today. And that, improbable as it may be, you should always try to work with a buddy. (or possibly pay a kid you can tolerate to stand watch in the remote location while you work on your impedance matching. whoops just kidding) Gotta note, tho--so many ghosts migrate from......Outside the machine!!!!

ohjustsomebodyelse (not verified) – Sat, 06/30/2007 – 19:23

Comment:  Agreed ...

It seems you do know more about BS than RF ... but not enough about either. I suggest some remedial education in electronics ... and start with inductance (something that can happen across wires running in parallel).

Captcha: NERDI .... nuff said

Captcha: GECBAG ... notice a trend here?

benezzell – Sun, 07/01/2007 – 00:10