Atomic I/O letters column #127
Originally published 2011, in Atomic: Maximum Power ComputingReprinted here February 8, 2012 Last modified 16-Jan-2015.
Feck!
I have this wooden "In" tray attached to my desk with double-sided tape. When I'm plugging a hard drive into a USB adapter cable, I put it in the tray, so it can't be knocked off the desk or pulled off by the cable.
The fecking roof fecking leaked over the fecking tray. All fecking night.
The fecking drive was 100% immersed in fecking dirty leak water. Said water is currently leaking back out of the drive as it sits out on the balcony in the sun.
I don't care about the USB adapter, so of course that still fecking works, after just swinging it around in the air to get water out of the plug.
And, the inevitable fecking punchline, there's stuff on that drive that I need and have not backed up. I know. I know. Shut up.
Is there any hope at all of getting any data off that drive ever again?
A
Answer:
Yes, you are almost certainly fecked. But what the feck, let's give it a go.
Hard drives, as you have now dramatically seen, are not sealed. They have a little pressure-equalisation vent with an air filter over it (and usually another filter, inside, for the air that whizzes around with the platters). The filter in your drowned drive probably worked OK as a water filter, so with any luck no particulates have gotten into the drive. But with just that one little vent hole, the drive will take a long time to dry out, unless you do something drastic to help. (It's surprising how much computer hardware can be rescued from fluid damage if you get rid of the fluid quickly enough.)
If it were my drive, I'd take the lid off it (you'll probably need a Torx "star" driver, or a near-enough hex Allen key, to remove the screws), put it in the sink, and pour methylated spirit over it to displace the water from nooks and crannies.
Next, dry out the lidless drive in as non-dusty a way as possible. While we're pushing the envelope, why not try the minimum possible heat in a fan-forced oven?
(The management accepts no responsibility if you set your alcohol-soaked drive on fire.)
Now pop the lid back on, plug the drive back in, and sit down, just in case it turns out to work and you, understandably, faint.
A drive that's had its lid off outside a dustless cleanroom, even if there was nothing wrong with it before, is a drive which should be dismantled for fridge-magnets and wall decorations after you quickly recover whatever data you can. Never mind dust - a smoke particle on a modern hard drive's platter is to the read/write heads like a bulldozer parked across a highway. A fingerprint on a platter is several times as tall as the flying height of the heads.
So I really, really wouldn't expect this to work for you. But it's an easy enough thing to try. It shouldn't make the drive much worse than it already is, either, so if the data's really valuable to you, you'll still be able to send the drive to an expensive recovery service.
Provided they don't just say "you're fecked" too, of course.
This window intentionally left blank
I've got a Compaq laptop and I thought I'd uninstalled all of the bundled crapware, but every time I boot some damn thing runs and makes a little square window with nothing in it. I've tried killing tasks in Task Manager, but I can't find what's doing this, or at least can't find what it is before I kill some task that's essential to Windows and have to reboot.
I also tried switching from Aero to old-Windows mode, and now I get the same empty square window with different things in the corners. Wunderbar.
I don't reboot that often so I don't really care if I can't stop the empty window from appearing. I just want to be able to close the thing. How?
Zach
Answer:
One possible cause for an annoyance like this is some sort of half-finished installer that isn't in the normal startup
lists, but tries, and fails, to finish setting some damn thing up every time you boot. Windows tries to prevent users
from accidentally stopping half-way through an install process that requires a reboot; aborting installers is, in
general, a bad idea. But sometimes the installer fails anyway.
I'm pleased to announce that one program can solve both Zach's problem, and this next one!
Once it goes black, it won't come back
If there's an open beta for a game, I'm in, even if it's an open-source multiplayer Ultima III (actually, that would be awesome).
I also like weird tweaked video drivers.
So I am pretty familiar with the Crash To Desktop. Nobody to blame but myself.
What I really hate, though, is when the game doesn't crash to desktop, but instead to a black screen with no mouse pointer. Now I have to ctrl-alt-del and open Task Manager and then use an invisible mouse pointer to click on the "this program has..." thing, or fumble the same invisible pointer to Task Manager and click the game and then End Process.
There must be a better way. You must know what it is. What is it?
CJ
Answer:
There's no perfect way to kill a system-monopolising foreground task like a game, when it falls over in the black-screen
way. But if you can zap the program by doing the pin-the-tail-on-the-Task-Manager routine, you'll probably also be
able to exit the program with "SuperF4".
When you press Ctrl-Alt-F4, SuperF4 kills the foreground program. If Windows still reckons that's the game, then that's what'll die.
SuperF4 also has an "xkill" mode, that terminates the program responsible for any window you click on - which should solve Zach's problem, above.
Even the standard mode of SuperF4 is a bit of a live grenade to have just sitting around in your taskbar, but it's easy to disable or quit SuperF4 when you don't feel a gangland program assassination is likely to be necessary.
If Zach would like to shoot for the moon and identify, rather than just kill, the process that's creating those stupid empty windows, Sysinternals' famous "Process Explorer" can ID the process that owns a window. Just click and drag the crosshair icon from the Process Explorer button-bar to any window you like. Then you can use another Sysinternals utility, "Autoruns", to turn over every system-startup rock and find where that process is coming from.
Rounder zeroes and straighter ones
My mate says that iPods sound better than most if not all other MP3 players, because they use a "better algorithm" to decode the music.
Is this true?
David
Answer:
Well, Apple's pet lossy compression format, AAC,
does sound better than MP3 at a given bit rate. There's also
Apple Lossless, which decodes to be bit-identical to the
original uncompressed PCM audio and so, all other things being equal, sounds exactly the same as the uncompressed
audio. (Or any other lossless compression, like FLAC or
Monkey's Audio.)
There's also some debate on the quality, or lack thereof, of the headphone output of different iPods. The unusual "push-pull" output of the original iPod Shuffle's headphone output ought to have sounded better than the normal "single-ended" headphone output of pretty much every other MP3 player ever made - but as usual in "audiophile" discussions, there are a lot more opinions being expressed than blinded tests being performed.
IPods definitely do not have a "better algorithm" for playing MP3s, though. They use commodity decoder chips - SigmaTels, Samsungs, maybe some others. None of these chips are Apple exclusives; they're all available in a range of players, including some very cheap and nasty ones. Not that this matters, because there's zero room for innovation in MP3 decoding algorithms.
There are lots of tricks you can do at the MP3 encoding stage, but in decoding you just take the bits and make a waveform out of them, end of story. Most MP3 decoders are "bitstream compliant", producing mathematically identical output, within the limits of rounding error.
Different MP3-decoder chips have different extra stuff built in - headphone-amp circuitry, screen-driver hardware, et cetera. But as far as decoding goes, they're either bitstream compliant and identical, or cut-down non-compliant chips for undemanding applications.
Shorter than mustard, longer than ham
I've been making backups since tape drives plugged into the floppy controller and took five million years to store 40Mb. Not once have I felt really confident about my backups. Floppies, tapes, CDs, DVDs; everything dies, unless you never need to restore from it. Who am I kidding, those ones die too.
Now I use Mozy to back up really vital data (and trust that they know what they're doing), but I've got terabytes of other stuff that I'd like to stick on cheap hard drives, then just put the drives in a safe place.
If I come back in ten years and try to restore these backups, will the drives be as hosed as every other backup media I've tried?
What's the shelf life of a hard drive?
Dan
Answer:
As I always find myself saying when people ask about drive reliability: Nobody knows. But probably a pretty long time.
Hard-drive technology has changed immensely over the years, but unplugged drives kept in a cool, dry place have always been likely to still work after several years. It's possible that the current ultra-high-density perpendicular-recording drives have some oddities that'll lose your data in less than five years, but it doesn't look likely. I'd put money on your data still being there in five years, and I wouldn't be surprised if the drives were still fine after ten.
For real security, though, you need to do three things.
One: Make more than one backup of anything really important. If you're backing up to hard drives, spread the risk by using different brands of drive.
Two: Do periodic test-restores. Even major corporations sometimes fail to do this, with predictably disastrous results. You want to discover dead backup media when it's annoying, not when it's catastrophic.
Three: Transfer your old backups to cheaper, bigger media in the future. Don't just back up new data to the $100 50Tb SSD you just bought from Tyrell Corporation (with other ones from Omni Consumer Products and Weyland-Yutani, as per point one); move the old backups onto the new media, too. This also gets around the problem of finding some way to plug a SATA drive from 2011 into the computer built into your brainstem in 2031.
(You can also, by the way, store small amounts {by modern standards} of digital data surprisingly easily on paper, which is a very stable medium.)
