winter's computers
found, loved, and nurtured for spring
my ugly geek secret

Saturday 16 June 2007

the big sleep

biglug wouldn't boot. The four AA's were long dead, leaking acid into the cheap, black plastic battery holder, my homegrown replacement for the original CMOS battery. It had lost all knowledge of what it was: what floppies, what hard drives, what monitors were attached to it. It didn't even know what time it was, convinced it was January 1st, 1980, flashback to big hair and FM synthesized music. New batteries would have to wait; I wanted to know what was on that hard drive.

Hours of wandering FTP sites in the lonely, dark corners of the world-wide-cobweb had turned up gold in the form of an ancient Compaq setup and diagnostics diskette image. Having dumped it to a dusty 720K disk — its label torn, words scratched out and over-written in an indecipherable history of recycled media — I brought biglug up and filled it in on what it was, what it could do (or more to the point, what it couldn't do), that the end of the millenium had come and gone and it was now an antique, sought more for nostalgia than number crunching. With the CMOS freshly updated, I ejected the diskette and rebooted. No dice.

Hard drive stiction. My old nemesis. "Never though I'd see you again," I croaked, the bitter memories rising like smoke from a sixty watt soldering iron. There was a time when a hard drive's read/write heads — gliding on Bernoulli's cushions of air — would land on the platters when power was cut, right on top of whatever data was being read at the time. Sometimes the heads would stick to the platters and the platters would refuse to spin up.

The first few times the client could finesse his way out of it, cycle the power repeatedly and maybe... just maybe... his drives would spin up and he could buy a little time to make backups, something he should have been doing as regularly as changing his underwear. Eventually the drive would stop cooperating and I'd be called in. I'd show up — often just after the client had changed his underwear — pull the drive out, hold it in one hand, and give it a quick twist. Typically that's all the convincing it would need.

I started down the path I'd taken so many times before, removing the covers, opening the card cage, releasing the drive cages. It was an old routine and biglug yielded willingly, though it shed pieces of its covers every time regardless of how much care I took.

I released the hard drive cage from the bottom as well, replacing the batteries while I was at it.

To protect the drive from stiction, it was common practice to "park" the heads in the "landing zone," an area of the hard drive with no data. This was before "voice coils," where the heads were attached to springs that would pull them back into the landing zone whenever the power was cut. Designed by dames, I reckon. A real man would manually run a utility that parked the heads before shutting down, taking his chances with power failures. After all, you always had your backup.

Even in the landing zone drives could experience stiction, though, as happened to biglug. I know I manually parked its heads before shutting it down last time.

Once I wrote a utility that parked a drive's heads on the other side of the disk from the landing zone in order to avoid stiction. I just partitioned the drive to skip the first and last cylinders so nothing important would be saved there, effectively creating my own landing zones. I wrote the utility in machine language using the DOS debug command, saving it to a .COM file. Then I bit the top off a bottle of whiskey, downed the entire contents in one go, and dug a bullet out of the back of my neck with a rusty steak knife dripping with iodine. Or something like that... I'm a little hazy on what happened after I wrote the .COM file.

A little twist, and with the boards plugged back in, the hard drive was spinning again.

There was something wrong, though. I needed to configure the CMOS again since I had swapped out the batteries, but the setup and diagnostic diskette wasn't booting. I checked my connections; everything looked good. The floppy was trying to boot but would just make a bunch of noise and fail. I immediately swapped out the floppy drive, worried but hopeful. It would be OK. biglug would pull through. We'd been through much worse than this.

The new drive was making the same noises. Controller problem then. Or motherboard... no, don't think like that. Get a hold of yourself.

I checked the connections again. The proprietary floppy controller wasn't fully seated. I cursed myself for not taking the appropriate care, for not checking twice before powering up. I reseated the controller but it was too late: biglug started with the same pathetic sounds.

This was it then.

I slowly reached over and placed my hand on the power switch, hesitated, then forced myself to toggle the switch and shut biglug off. One last look at the CRT, the words "601 - floppy error" fading slowly with the persistence of the old green phosphors, and I locked the keyboard back in place one last time over the dark screen.

It was the big sleep for biglug.

No comments: