Question how XP can repair itself

[jimdagys]

I was scanning a picture, and as the jpg was being written to the hard drive, the power went off. Later, when I tried to start the computer, I just got a screen something like "insert boot disk". I turned it off and on, but no change. Somebody looked at the computer and did a few tests on the boot screen and he said the hard drive was probably destroyed. A few hours later I tried to start the computer, this time the "check disk" came on. It seemed to finish the check disk normally, then Windows started. It seemed to boot up OK. I shut the computer down and restarted it and everything seemed OK.
My question is: How can Windows seemingly repair itself in this situation?

[transporterfan]

http://support.microsoft.com/kb/831426

[mike20021969]

I'm interested in why CHKDSK did not run when jimdagys started computer first time after power went off.

It seems it took at least 3 tries to boot before CHKDSK ran.

[jimdagys]

It turns out that transporterfan is a robot and not a person. It just scanned the key words in my above question and then gave an answer, which has nothing to do with my question. So I am still really curious about how XP could repair itself. Checkdisk did not run after the first several attempts at starting the computer. Only the bios screen came up. I'm thinking XP must have some critical data written in two different places on the disk, and therefore was able to repair itself. But that still doesn't explain why the computer initially wouldn't boot, but 2 hours later booted up.

[jman98]

Your assumption that Windows somehow "repaired itself" is almost certainly wrong. I'd put the odds of that really being what happened at 1% at best. Others will have to provide a mechanical explanation for why the drive eventually responded.

[TimA-C]

XP (and later versions of Windows) can repair itself by rolling back drivers/files and/or going back to earlier Restore points, but I don't think that's what's happened here. What you've described sounds more like a capacitor problem either in the power supply or on the motherboard. The "A few hours later" possibly gave the misbehaving capacitor long enough to discharge itself.

If it happens again try unplugging the mains lead and pressing & holding the power button of the front of the PC for up to 20 seconds before reconnecting and trying to turn on in the usual way. Sometimes unplugging and plugging back in the 20/24 pin power connector on the motherboard will cure it as well (at least for a while) but I wouldn't recommend doing this unless you're comfortable with opening up a PC and working safely on it.

[jman98]

XP (and later versions of Windows) can repair itself by rolling back drivers/files and/or going back to earlier Restore points, but I don't think that's what's happened here.

Yes, but they don't do this without human intervention telling them to do so. We do agree that this isn't what happened and your explanation seems likely.

[transporterfan]

It turns out that transporterfan is a robot and not a person. It just scanned the key words in my above question and then gave an answer, which has nothing to do with my question.

Fair enough. I thought you were just asking how CHKDSK can run of it's own accord. For what it's worth (and I don't have access to your system to check), I think you have a bad CMOS battery, or loose connection or faulty connector to the battery. The fact is there are a number of reasons why a disc wont boot/initialise but the CMOS checksum is as good a guess as any.

If the CMOS checksum is considered corrupt or null the system wont start. Let it rest a while between boots and, maybe, there is enough charge in the battery to sum once and, as a result, the boot sector is read and CHKDSK is triggered because the part saved image file caused a discrepancy in the FAT.

At this point the computer boots and (seemingly miraculously) the system repairs itself.

Truth is it's anyone's guess. Just be glad you got back in.

[hello_hello]

The way I understand it (someone correct me if I'm wrong), a power outage, even while a hard drive is reading/writing, shouldn't hurt the hard drive. It might cause data loss, or mess with the file system etc if the drive's writing data at the time, but it shouldn't hurt the drive itself. When the power is lost the heads are returned to their "park" position by a spring and the platters keep spinning long enough to maintain the "cushion of air" which keeps the heads from touching them. I don't think hard drives shut down any differently when you shut down a PC "properly".

I wonder if maybe the power outage caused a problem with the BIOS and after a few attempts at starting which failed, the BIOS reverted to a "last known good" state, or maybe it's default settings. I'm not sure how likely that is but maybe it'd explain a few false starts where the BIOS couldn't see the hard drive,. When it finally did and booted from it, Windows automatically ran CHKDSK because it hadn't been shut down properly.
"Somebody looked at the computer and did a few tests on the boot screen and he said the hard drive was probably destroyed" sounds to me like the hard drive wasn't showing up in the BIOS for some reason.

[fritzi93]

From my experience, I'd investigate 3 possibilities, one of which has not been mentioned yet:

1) Iffy SATA cable connection. Yeah, it seems unlikely to have a drive disappear, then re-appear upon reboot without doing anything else, but I've seen it a few times. And I recall chkdsk running at least once in that scenario. Re-seating the cable fixed it temporarily; in the end the cable always had to be replaced. Get the locking ones.

2) BIOS settings got screwed somehow by a power outage. Jumpering the pins (on the mobo) for 10 seconds or so and discharging the capacitors should fix that.

3) Dying capacitor(s). Here's where having a spare power supply is really useful, as you can pin down whether it's on the PSU or mobo.

[Constant Gardener]

Had an XP system that was working when lightning took out the power. When power returned, computer would not boot. Tried several times. After a couple of hours, I tried again, and the system came up perfectly. I have no idea what happened, since it was a remote site and I was manipulating it over a phone line. (Of course the phone and power went through an APC power bar, which was partly fried and subsequently replaced.)

[Bjs]

You'd be wise to invest in a ups ... it will at least if fully charged, be used to confirm if issue is at pc side or the power supply side.

Highly likely the system board need time to reset from the issue ... asus motherboards with dual bios do the same thing ... and they can drive inexperienced technicians insane figuring whats wrong with the system.

Rule of thumb:

Start from point A, not point C ... If you don't know what the quality or stability of the supply source is then your already well behind.

[Nelson37]

Windows did not repair itself, it never booted until the chkdsk operation happened.

The "boot screen" the other person looked at, likely the BIOS menu and likely, the hard drive was not detected. This would be very important to find out.

The error msg means either there is no hard drive detected, or it is not bootable.

The BIOS likely has no major problem, the POST test completed and the board attempted to boot.

The problem is with the drive, data or power to the drive, or the dreve controller. If you had listened to, or felt, the drive at power up, then we would know if it simply was not spinning, or slow to come up to speed. Warm re-boot can sometimes detect this.

[deadrats]

windows has been able to repair itself since xp and later version were even better at it. i know for a fact that windows logs failed attempts to boot and i also know that it checks said for said log each time it boots, i think a fair assumption is that it can use the log entries to correct whatever caused it to fail to boot.

i used to have an old pc that ran vista that every once in a while it would fail to boot for no apparent reason, either it would hang, windows would fail to load, sometimes i would get an error message but sure enough if you tried again a couple of time it would boot up and everything would be fine. i forget where the log file is but i remember tracking down in vista and trying to poke around in it but it was so long ago i forget what info it contained.

btw, windows is not the only OS with self healing properties, Solaris had this feature years ago as did server versions of windows.

[Nelson37]

There is no log file created when you get "insert boot disk". The OS has never even begun to boot.

Nothing whatsoever changed concerning the OS until the last "successful" boot when CHKDSK ran, and "repaired" that particular problem, which was minor.

The MAJOR problem was that the drive was either not detected at all, or was detected as non-bootable. This is a mechanical or electrical problem not involving the OS in any way. THIS is the "mysterious problem" that somehow "self-repaired".

[gonca]

The way I understand it (someone correct me if I'm wrong), a power outage, even while a hard drive is reading/writing, shouldn't hurt the hard drive. It might cause data loss, or mess with the file system etc if the drive's writing data at the time, but it shouldn't hurt the drive itself. When the power is lost the heads are returned to their "park" position by a spring and the platters keep spinning long enough to maintain the "cushion of air" which keeps the heads from touching them. I don't think hard drives shut down any differently when you shut down a PC "properly".

I wonder if maybe the power outage caused a problem with the BIOS and after a few attempts at starting which failed, the BIOS reverted to a "last known good" state, or maybe it's default settings. I'm not sure how likely that is but maybe it'd explain a few false starts where the BIOS couldn't see the hard drive,. When it finally did and booted from it, Windows automatically ran CHKDSK because it hadn't been shut down properly.
"Somebody looked at the computer and did a few tests on the boot screen and he said the hard drive was probably destroyed" sounds to me like the hard drive wasn't showing up in the BIOS for some reason.

A lot of mobos, nowadays, will reset a corrupted bios back to default if the power is shut off for 20 seconds or so. also a lot of mobo's will automatically reset the bios to default on a failed boot,post

[hello_hello]

A lot of mobos, nowadays, will reset a corrupted bios back to default if the power is shut off for 20 seconds or so. also a lot of mobo's will automatically reset the bios to default on a failed boot,post

This motherboard's BIOS certainly isn't afraid to reset itself. The main problem is it doesn't inform the user of it's newly reset state, although I vaguely remember reading somewhere Gigabyte have rectified that for their newer motherboards.

[gonca]

They do it automatically, however it should inform the user through an error message of some type.

[jimdagys]

I asked the man who looked at the computer and he said that the BIOS didn't detect the hard drive. My understanding is that he said it was like there was no hard drive. Somebody here mentioned a bad connection, but I have never before seen any evidence of bad connection with this computer. All I know is that I've been using the computer for several days since that happened and there has been no problem. The computer seems to work perfectly. So it wasn't XP that repaired itself. I guess it was the hard drive, the BIOS, or hard drive controller that somehow got working again.

[hello_hello]

It sounds most likely that the BIOS eventually reset itself, although the man who looked at the computer for you likely reset it manually to see if the hard drive would then appear. Or maybe he didn't....