Yesterday I had an emergency call from one of my few remaining SCO customers. The system had crashed and the monitor was showing "No input".
There are a couple of possibilities here. One would be that it's just the video on the motherboard that has gone south and everything else is fine. If that were the case, their network and terminals would be working, but they were not, so obviously more was wrong.
I jumped in my car and headed toward them while still on the phone. This is a retail store; they need this system. With that in mind, they had purchased two identical machines originally, so of course our thoughts turned to that. We could restore backups to the twin machine, but that would take some time and apparently the machine had crashed before doing the backup last night. Moving the hard drive seemed like the quickest way to get back in business, so they started on that while I drove.
These boxes have drives in easy to remove cages, so it was only a few minutes later that they called back saying that the system had booted and was asking questions about a dump image.
"There may be a system dump memory image in the swap device. Do you want to save it?". Well, no, we don't. I don't really care why this system crashed. Obviously the mother board lost its video; who knows what else happened? More importantly, who cares - nobody is going to analyze the dump image to see what
happened. I told him to answer "N" and to answer "Y" to the next question about deleting it. Then, of course, he got "fsstat: root filesystem needs checking
OK to check the root filesystem (/dev/root) (y/n)? "
"I usually say No", he said.
Huh? No, no no, you have to say "Yes".
"But it takes a long time", he protested.
Still driving, I explained WHY he needs to let fsck look through his drive. I'm quite sure we have had the exact conversation at some point in the past, but he must have forgotten. Fortunately, fsck ran through without any serious problems and the system continued to boot. Eventually he had a login prompt.
The console was working and so were the dumb terminals, but TCP/IP was not. I wondered if we really did have truly identical machines - I was pretty sure they were, but why wouldn't the network have come up?
I thought about that as I drove. Could the NIC be in a different slot? If so, I could just lead him through running netconfig, but I really remembered these as truly identical machines. He was mostly working, I had only another fifteen minutes of driving - I decided to not have him do any more. He could limp along until I got there.
When I arrived, I wasn't surprised to see that the machines really were twins. So why hadn't TCP/IP come up? I checked "ifconfig -a" and it HAD come up. It just wasn't working. Peeking around the back, I saw no lights at all on the card. Dead card? Maybe, but usually that would cause it not to be seen at all. Maybe I should just unplug it and check the patch cord? I put my fingers on it and the lights came on..
Hmmm.. I had barely touched it - it wasn't loose, wasn't unplugged. The cover was still off the machine and I could see that the card was securely screwed down and seated. What the heck?
The owner started reaching toward the wire. I stopped him.
"Look, you are working now. Yes, there is something screwy with this card or the patch cord. Why not deal with that tonight after the store is closed?"
He agreed with that. "So can you fix the other machine?", he asked.
Aaargh. Yes, but.. it's time to move on. SCO is nearly bankrupt (they apparently got bought again but that can't mean anything). His app runs under Linux, the hardware is five years old; it's time to dump SCO and move to Linux.
Time? It's way past time. If he's really strapped, he could get a new motherboard and use this old hardware, but I'd rather see him just buy new - hardware is cheap.
He agreed, so I had him call the app people to get that going. Another SCO bites the dust.
Got something to add? Send me email.
Increase ad revenue 50-250% with Ezoic
More Articles by Anthony Lawrence
Find me on Google+
© 2009-11-06 Anthony Lawrence