Slashdot Mirror
Computer Crash Reactions Examined
dankinit writes "MSNBC has an amusing story about research showing how people react to computer crashes and losing data. Among the numbers, 7% of those surveyed hit the computer, 13% yell at first, and another 13% try to "sweet-talk" their computer. The article also has results from a study done at the Univ. of Maryland. In that study, "One restaurant manager who was so upset with his laptop that he threw it into deep fryer. That destroyed the laptop ... and deep fryer, too.""
It disappears; the system is completely halted. You only continue to see the screen image because the video hardware continues to read the contents of VRAM and send it out the display port.
Peripherals that use hardware passthroughs can have amusing responses to a system crash... When an old machine that had a TV card froze up, the video and audio would continue playing (the video was put onscreen as a hardware overlay on a key color, and the audio was handled by the PCI card). The audio would continue to play as the system rebooted, only stopping when the computer got around to loading the TV card's driver which reinitialized the hardware.
The PC God cannot be appeased by poetry. He requires human sacrifices.
Or pr0n.
You might look at Unison, which runs under Linux and Windows. It has been working great with our Linux laptops (used to sync up user home directories). It is nice because it allows work to happen in two places, and then when you sync up it copies stuff everywhere it should go (and gives you an opportunity to manage conflicts). And I agree on the whole backup deal; I am planning on another machine soon with sufficient diskspace to mirror all the data I care about, plus 2 removeable drives, the most current of which I can store at work.
I Am My Own Worst Enemy
It depends on a lot of things. If the crash is due to a hardware failure, the machine might hang inside a device driver, waiting for an IO acknowledgement that will never come. Sometimes the OS deliberately halts the machine, such as when it receives a double-fault inside kernel space. Back in the DOS days, a bad address (usually a result of a buffer overflow) would shoot the machine into la-la land, executing random garbage until it either reached a HLT instruction or got stuck in a loop.
Again, in the days of DOS, it was tons of fun when you accidentally used a far NULL pointer, since that meant overwriting the interrupt vector table -- the next time a timer interrupt came in (18.2 times a second, to be precise) the CPU would happily fly off into dreamworld.
On most MODERN operating systems, however, the system deliberately halts itself as soon as it realizes that something is terribly wrong. The risks of executing random code are simply too great.
Yes, it's called Bookmarks Synchronizer and is a Firefox extension
Rik
However, do not use this sort of thing for backups!!
If files get deleted by accident and you don't notice until after the backup, BOTH copies of the data will be bad. This is just replication, which itself is very good for somethings. Not backups though.
I use a tool called RIBS. It uses rsync to create incremental backups across the network. You get hourly.0, hourly.1 etc directories, each with a hard-linked snapshot of the backup as it was at that time. These pan off into weekly, then monthly. Personally I dropped the hourly entirely, I just do daily. I do this off an IDE disk onto an IDE disk, no RAID or anything fancy required. Sure, it's a little extra work should one of the drives pack in (no RAID redundancy, maybe one day perhaps), but it's worth it for the cronological snapshots. I even backup files like MS Outlook *.pst on my Windows box, so should it get corrupted, I don't care.
Oh, did I say I'm backing up 120GB of data onto a P90 with 16meg of ram? Not bad for old junk!
Only the deltas are transmitted with it being rsync. Highly recommended, knowing you can restore ANY file means I haven't renamed to *.bak in a very long time!
I wonder if different OS crashes induce different responses?
Yes, of course. When Linux crashes on me, I start looking for hardware problems (and most often than not, I find them!). When Windows crashes, I just think "well, it's Windows", reboot, and continue. When Mac OS X crashes... I don't know, I think I've never seen it crash. I use my iBook far less than my PC though.
My website
>>Linux is *hard* to crash Unfortunately if you use a program called root (a hated high energy physics histograming package, the name actually is a very good one because the fact that is linuxed based and called root basically says it all) you can easily crash linux. I find this one of the more impressive things about the program.