A local business recently retired its elderly Xeon server, replacing it with something a little more powerful; the Xeon was repurposed as a normal desktop PC, with a new copy of XP Pro and all. The guy who was to get it asked us to install a sound card, so we did that.
And then it stopped working, saying we’d triggered the chassis intrusion alarm. On most computers, you’d go “oh, okay, that’s cool”, hit F1, and go about your business.
Not so with a server based on an ASUS NCLV-DA. This board yanks on the air-raid siren, barks at its own barking, and thumbs its nose at you if you take the side off the case. It refuses to boot until you reset the CMOS, which of course wipes out the clock and other settings.
At least one other unlucky tech had come across this problem before me; someone’d stickytaped the switch down, so it wouldn’t trigger if the case side came off. This had served well until today, when the tape let go and the motherboard went into panic mode in front of me.
I had a more permanent solution up my sleeve: I stuck a jumper across the intrusion sensor pins on the board, effectively closing the circuit and cutting the little switch and its ever-so-thin wires out of the loop entirely.
This computer had been hanging frequently, after killing one PSU two months ago; it was only while swapping out RAM during testing that I noticed the damage here.
I shudder to think what might’ve happened if I’d left it memtesting overnight.
My exact words were “Hmm, those wires feel… what the f… jesus, the… wow!”.
In fixing a computer, you must narrow the problem down to being one of two types: hardware and software.
A software problem is a genuine bug in a program, where clicking something ordinary produces an unexpected and often inexplicable result, like incorrect output or a funny error message and a crash. These problems are often fixed by updating the software to a newer version, or replacing it with something equivalent.
A hardware problem is a faulty video card, or a failing hard drive, which often mainfest themselves in software-problem-like ways (programs glitch or crash) as far as the user is concerned, but are part of an overall pattern that can point to hardware. A dying hard drive can cause a computer to run slowly, programs to crash on startup, Windows not to start up at all.
It’s important to differentiate between the two, and recognise that one may look like the other. One can also cause the other; bad memory can corrupt data to the point where Windows doesn’t start, and you’ll still have to reinstall after replacing the faulty RAM.
There are also, I’ve come to realise, two types of computer users.
The first type will say they know just enough about computers to get themselves into trouble. They are often wrong about this, and their computers will be clean as a whistle, with very few desktop icons, they’ll know how to email spreadsheets to people, they’ll ask intelligent questions (and be honest about the dumb ones) and they’ll eagerly listen to every word you say with deep fascination and concentration. These are my favourite customers.
The second type are more than happy to admit they tinker with their PCs, often without knowing what’s happening or how it can affect things. Then, when they show you the computer you built them and it’s got scratches and cat hair all over it, they’ll arrogantly accuse you of selling them a useless antivirus program and that they clearly have a massive infestation of some sort. Then you’ll discover they’ve moved the folders for those apps out of C:\Program Files\ and into C:\Documents and Settings\Greg\Desktop\Spies and Nasties\, and argue til they’re blue in the face it’s not their fault Spybot doesn’t load anymore.
This second category of user brings me to my point: There’s a terrifying, third kind of computer problem you can encounter. Console modders will be familiar with the concept of a semi-bricked device – not quite a dead write-off, not quite alive enough to be usable, hovering between this world and that in a curious limbo state.
They don’t sleep, they don’t communicate, they walk aimlessly forever unless you cut their heads off, and they can’t be brought back. I’m talking about zombies. There’s nothing you can do except to step back, nuke from orbit, and start all over again.
One lady dropped her computer about a foot onto a concrete floor. No cracks in the case, no cracks in the motherboard, no bad sectors on the hard drive, nothing verifiably wrong with it, but it… ran… slowly. Installing XP would take three hours. It’d take minutes to get to the desktop. Everything would work, but at a snail’s pace. We sold her a new computer.
One old guy did something to Device Manager or the drivers, and his computer wouldn’t recognise PS/2 mice. USB mice were fine, PS/2 keyboards were fine, but that PS/2 mouse was an unknown device and nothing could ever be made of it. There was no fix – this was not a problem that sensible, normal people have with their computers, so nobody had ever written a fix for it. We reinstalled Windows and told him not to do it again.
Another guy ordered a computer with four gigs of RAM last week – specifically, two twinpacks of 1GB DDR2 made by G-skill. Because the days of random memory incompatibilities are long over, we agreed, and he brought it back a week later because it bluescreened five times a day and it was essentially unusable. We narrowed it down to the RAM, and G-skill told us we should’ve bought a 2x2GB kit instead of using four 1GB sticks to ensure compatibility. We put the Gskill RAM into other computers separately (they work fine when there’s not four of them together) and sold the guy RAM of a less-picky brand.
When a printer’s quick setup guide tells you not to plug the printer into the computer until you’ve installed the software and it instructs you to do so, do what it bloody well says, because if you plug the printer in first and Windows starts in with the “oh crap, where’s the software for this thing?” dance, you’ll need someone with a steady hand, deep knowledge of Windows, and a twinkle of danger in their eye to make things right afterward. I don’t know how anybody can be proud to sell a consumer-targeted inkjet made by HP, because troubleshooting those things is like trying to debate philosophy while being repeatedly hit in the face with an angry wombat.
This was a brand new P300 out of the box. Customer wanted XP on it, so we re-imaged it with the supplied recovery CDs and took it back from Vista to XP.
And wireless networking stopped working. It has an Intel Wifi Link 5100 card, which is definitely new enough, but 99% of the time it could see no wireless networks, and the few moments it could detect them, it couldn’t connect to them.
The strangest thing was the reported CPU usage – Task Manager reported the SYSTEM process to be using 20-40% CPU, all the time, yet the System Idle process was sitting at 99% as well.
Yes, I realise those numbers don’t add up. That’s what it was doing. I wish I’d had time to get a screenshot.
Same as with the Acer laptop a couple of weeks ago, I loaded Process Explorer and found the same old Deferred Procedure Call hog. Like the Acer, disabling a certain device made the problem go away; in the case of the Toshiba, it was the wireless card.
Toshiba Australia told me to re-image it with the recovery CDs again, and if the problem still occurred, to contact our supplier for warranty. Only after going through that process again and calling our distributor were we told there was a known problem with the P300s running Windows XP involving wireless, and the fix for it is updating the BIOS to the latest version.
Toshiba’s AU page is here, and it uses such a weird URL structure I don’t think I can give a direct link to the download page. It’s not that hard though – Support on the left, Drivers, BIOS, Utilities, Software in the middle, Notebooks -> Satellite Pro -> P300-(PSPCxA) depending on your exact model – first two links are Windows-based BIOS flashers, depending on whether you’re using 32 or 64 bit Windows.
I updated the BIOS, handed it to the customer, everything’s all good now. Don’t forget hardware-fault-looking problems on brand new hardware might be fixable with a BIOS update – even and especially on laptops where the problem device isn’t actually a component of the motherboard!