Why must you shutdown, XP?
May 21, 2006 6:34 PM Subscribe
Why must XP insist so infernally on rebooting?
My wife has an IBM NetVista (2292 CBU). It's about four years old. It is running Windows XP Pro, SP 2. Lately, it has decided to reboot with little reason or warning. At its worst moments, it will go into a rebooting cycle where the power supply will whirr to life, the IBM splash will come up, then it will power down. This repeats. Other times, when powered up from the button, it will show the IBM splash, the WinXP splash, then power down. Most of the time, it will simply reboot, usually after popping up the WinXP shutdown dialogue box. She can hit cancel, only to have it return soon after. And that second time, hitting cancel only gets rid of the dialogue, but not the eventual reboot.
So far, this is what I've been able to discover and/or investigate:
- Certain models in this IBM NetVista line were notorious for having/developing . The symptoms for that problem align, but an examination of the board reveals none of the tell-tale signs of capacitor leakage.
- In the XP system events log, there was a regular error upon bootup (confirmed by looking at the boot log) involving a missing ASPI driver. While rare, it appears that a malfunctioning ASPI32.sys can cause hangups. This driver has been reinstalled, the error no longer appears.
- When it would reboot, the fan on the power supply would audibly increase in speed. The stock power supply on an IBM NetVista is 185w. Since we've installed a better video card and a second harddrive, I thought that the power supply might be over-taxed, so the stock has been replaced with a new 300w. This seemed to stop the reboots, but only for a week.
- The IBM NetVista has a multi-function keyboard with a "standby key." Just to make sure, I've gone into msconfig and removed that app from the startup. This seems to have had little to no effect.
- Techie friends have been asking me to supply them with the minidump file, but since this appears to be a system-spawned reboot, there is no real crash to record and then decode.
So ... I'm reaching the ends of several ropes. Sometimes, the machine will run for six or seven hours. Other times, it throws the shutdown dialogue in under ten minutes. I'm looking for any possible ideas or avenues that I've not considered, or any similar experiences. Fire at will.
My wife has an IBM NetVista (2292 CBU). It's about four years old. It is running Windows XP Pro, SP 2. Lately, it has decided to reboot with little reason or warning. At its worst moments, it will go into a rebooting cycle where the power supply will whirr to life, the IBM splash will come up, then it will power down. This repeats. Other times, when powered up from the button, it will show the IBM splash, the WinXP splash, then power down. Most of the time, it will simply reboot, usually after popping up the WinXP shutdown dialogue box. She can hit cancel, only to have it return soon after. And that second time, hitting cancel only gets rid of the dialogue, but not the eventual reboot.
So far, this is what I've been able to discover and/or investigate:
- Certain models in this IBM NetVista line were notorious for having/developing . The symptoms for that problem align, but an examination of the board reveals none of the tell-tale signs of capacitor leakage.
- In the XP system events log, there was a regular error upon bootup (confirmed by looking at the boot log) involving a missing ASPI driver. While rare, it appears that a malfunctioning ASPI32.sys can cause hangups. This driver has been reinstalled, the error no longer appears.
- When it would reboot, the fan on the power supply would audibly increase in speed. The stock power supply on an IBM NetVista is 185w. Since we've installed a better video card and a second harddrive, I thought that the power supply might be over-taxed, so the stock has been replaced with a new 300w. This seemed to stop the reboots, but only for a week.
- The IBM NetVista has a multi-function keyboard with a "standby key." Just to make sure, I've gone into msconfig and removed that app from the startup. This seems to have had little to no effect.
- Techie friends have been asking me to supply them with the minidump file, but since this appears to be a system-spawned reboot, there is no real crash to record and then decode.
So ... I'm reaching the ends of several ropes. Sometimes, the machine will run for six or seven hours. Other times, it throws the shutdown dialogue in under ten minutes. I'm looking for any possible ideas or avenues that I've not considered, or any similar experiences. Fire at will.
This definitely feels like a hardware issue. I want to say bad power supply, or a heat issue, but it might be something worse than that (like broken mobo/bad memory).
But, just in case, try running in safe mode for a bit, and see how long the box will stay alive. If safemode is okay, then it must be a software issue. To diagnose heat problems, try running the computer with the case open for a period of time. Listen for fans whirring irregularly (slowing down, speeding up, etc). If your CPU fan is dying, that might be it.
And the computer is plugged into a surge protector, right?
posted by Drunken_munky at 6:58 PM on May 21, 2006
But, just in case, try running in safe mode for a bit, and see how long the box will stay alive. If safemode is okay, then it must be a software issue. To diagnose heat problems, try running the computer with the case open for a period of time. Listen for fans whirring irregularly (slowing down, speeding up, etc). If your CPU fan is dying, that might be it.
And the computer is plugged into a surge protector, right?
posted by Drunken_munky at 6:58 PM on May 21, 2006
The wife here.
1) We're not seeing any errors w/r/t drivers other than the one mentioned above.
2) Powersupply is already replaced. Old one was overheating the box a bit, but temperature montiors after the new supply was installed showed both HD's and everything else at a respectable temperature.
3) The fan works. We ran the box open for a couple of days already. The over-extended powersupply was too warm and hella loud. The new one is whirring along pleasantly.
4) Motherboard is, upon visual inspection, fine, with none of the classic capacitor symptoms. I am a little afraid that the over-warm powersupply might have baked it a bit, though.
5) Yes, it's plugged into a surge protector.
We will try both memtest and safemode. Thanks!
posted by Medieval Maven at 7:08 PM on May 21, 2006
1) We're not seeing any errors w/r/t drivers other than the one mentioned above.
2) Powersupply is already replaced. Old one was overheating the box a bit, but temperature montiors after the new supply was installed showed both HD's and everything else at a respectable temperature.
3) The fan works. We ran the box open for a couple of days already. The over-extended powersupply was too warm and hella loud. The new one is whirring along pleasantly.
4) Motherboard is, upon visual inspection, fine, with none of the classic capacitor symptoms. I am a little afraid that the over-warm powersupply might have baked it a bit, though.
5) Yes, it's plugged into a surge protector.
We will try both memtest and safemode. Thanks!
posted by Medieval Maven at 7:08 PM on May 21, 2006
In my experience, spontaneous rebooting has been caused by either CPU overheating or memory going bad. This usually manifests as weird driver issues.
The original poster is in metro Atlanta, where we are experiencing the warmest weather of the year (mid to upper 80s). I toughed it out without turning the AC on today but -lo!- my wireless router starting getting flaky and I think it was overheating. Maybe the same thing was happened with the OP's PC. More to the point, a flaky problem related to overheating was made worse by warm weather.
Either way, these things are usually due to overheating or bad RAM. Check your fans (you have 2 or 3 of them) and run memtest overnight.
posted by intermod at 7:12 PM on May 21, 2006
The original poster is in metro Atlanta, where we are experiencing the warmest weather of the year (mid to upper 80s). I toughed it out without turning the AC on today but -lo!- my wireless router starting getting flaky and I think it was overheating. Maybe the same thing was happened with the OP's PC. More to the point, a flaky problem related to overheating was made worse by warm weather.
Either way, these things are usually due to overheating or bad RAM. Check your fans (you have 2 or 3 of them) and run memtest overnight.
posted by intermod at 7:12 PM on May 21, 2006
Response by poster: Running MemTest now ... actually, running Memtest86+ ... I'm not really sure just what I'm looking for, to be honest. If an error is found, I think the message is reported in hex. None have appeared, but if one does ... is there a decoder ring available?
posted by grabbingsand at 7:21 PM on May 21, 2006
posted by grabbingsand at 7:21 PM on May 21, 2006
It might help to disconnect anything you've got connected to the machine, especially USB devices. What's happening sounds to me like the power supply crowbarring.
I agree with everyone else; this isn't a software issue. This is some sort of hardware failure. It could definitely be heat related.
posted by Steven C. Den Beste at 7:44 PM on May 21, 2006
I agree with everyone else; this isn't a software issue. This is some sort of hardware failure. It could definitely be heat related.
posted by Steven C. Den Beste at 7:44 PM on May 21, 2006
The fact that you are getting the "WinXP Shutdown" dialog means that your BIOS has seen a problem - most likely overheating or bad power - and purposely invoked the shutdown. This should leave a message in your event log - is it not?
Unless, of course, it's happening on boot up, before the Event Manager has loaded. But if you got that AFTER you were already running windows, there should be an event in the event log.
posted by Dunwitty at 7:46 PM on May 21, 2006
Unless, of course, it's happening on boot up, before the Event Manager has loaded. But if you got that AFTER you were already running windows, there should be an event in the event log.
posted by Dunwitty at 7:46 PM on May 21, 2006
Depending on how confident you are, you should try stripping the system down completely and booting from a live CD like knoppix or BartPE. Then add a component and boot, add a component and boot, until it is fully rebuilt. If a part has failed outright you will know exactly which one right away. If the problem is intermittent you can chase your own tail doing this, but it is still the right thing..
The visual signs of Capacitor Plague can be very subtle. Also, fixing a system by power supply replacement can be an early indicator capacitors which are begriming to deteriorate. So, even though you mention it, it might be worth checking again..
On the other hand, capacitor plague is my favorite suggestion for all seemingly hardware related computer problems, so..
If an error is found, I think the message is reported in hex. None have appeared, but if one does ... is there a decoder ring available?
If you get a few isolated errors, you have a bad DIMM. If every single bit shows an error, the motherboard to DIMM interface is wonky.
I've had systems that showed errors with DIMMs in one order, but worked fine with the same DIMMs in another order:
bitr0t: How many hours have you spent debugging this problem, and how many do you intend to spend looking for a solution?
Indeed. But, shopping for a new computer can take as long as repairing this one.. There are no free lunches.
posted by Chuckles at 8:05 PM on May 21, 2006
The visual signs of Capacitor Plague can be very subtle. Also, fixing a system by power supply replacement can be an early indicator capacitors which are begriming to deteriorate. So, even though you mention it, it might be worth checking again..
Bulging or domed tops, no matter how slight, are a sure sign the capacitor is faulty.Emphasis mine.
On the other hand, capacitor plague is my favorite suggestion for all seemingly hardware related computer problems, so..
If an error is found, I think the message is reported in hex. None have appeared, but if one does ... is there a decoder ring available?
If you get a few isolated errors, you have a bad DIMM. If every single bit shows an error, the motherboard to DIMM interface is wonky.
I've had systems that showed errors with DIMMs in one order, but worked fine with the same DIMMs in another order:
DIMM A in slot 1failed, but
DIMM B in slot 2
DIMM C in slot 3
DIMM A in slot 1worked fine. So, persistence and tinkering can be worthwhile.
DIMM C in slot 2
DIMM B in slot 3
bitr0t: How many hours have you spent debugging this problem, and how many do you intend to spend looking for a solution?
Indeed. But, shopping for a new computer can take as long as repairing this one.. There are no free lunches.
posted by Chuckles at 8:05 PM on May 21, 2006
Response by poster: Unless you are unemployed and have ready access to cheap parts, fixing a four year old computer is probably not the best use of your time and money.
Possibly, but this computer is a resource that still has value. Replacing it will not come cheap, so if there is an avenue of repair that falls within an acceptable margin of expense, I see no harm in following it.
This should leave a message in your event log - is it not?
It is not.
posted by grabbingsand at 8:08 PM on May 21, 2006
Possibly, but this computer is a resource that still has value. Replacing it will not come cheap, so if there is an avenue of repair that falls within an acceptable margin of expense, I see no harm in following it.
This should leave a message in your event log - is it not?
It is not.
posted by grabbingsand at 8:08 PM on May 21, 2006
"- When it would reboot, the fan on the power supply would audibly increase in speed. The stock power supply on an IBM NetVista is 185w. Since we've installed a better video card and a second harddrive, I thought that the power supply might be over-taxed, so the stock has been replaced with a new 300w. This seemed to stop the reboots, but only for a week."
I think you might need more than 300W, particularly if you've added an AGP card. 400W is probably more than enough. Something to consider is how many amps each rail needs, particularly the 12V rail. http://www.dansdata.com/psus.htm http://www.selfseo.com/story-13011.php It's possible to get a decent 400W power supply for around $40-50. Maybe try to run the machine stripped down, with just memory, CPU and HDD. If it runs fine, then you may be able to isolate the problem by installing hardware one piece at a time, rebooting every time and running for a while to see what happens. If it starts giving problems when you install something, it could indicate the hardware itself is bad, or you don't have enough power for it.
Other than that, like other people said, maybe bad RAM, overheating, or a fried motherboard from the last (inadequate) power supply.
posted by krinklyfig at 8:29 PM on May 21, 2006
I think you might need more than 300W, particularly if you've added an AGP card. 400W is probably more than enough. Something to consider is how many amps each rail needs, particularly the 12V rail. http://www.dansdata.com/psus.htm http://www.selfseo.com/story-13011.php It's possible to get a decent 400W power supply for around $40-50. Maybe try to run the machine stripped down, with just memory, CPU and HDD. If it runs fine, then you may be able to isolate the problem by installing hardware one piece at a time, rebooting every time and running for a while to see what happens. If it starts giving problems when you install something, it could indicate the hardware itself is bad, or you don't have enough power for it.
Other than that, like other people said, maybe bad RAM, overheating, or a fried motherboard from the last (inadequate) power supply.
posted by krinklyfig at 8:29 PM on May 21, 2006
While everyone else is calling out the hardware, I'm for doing some more hunting around for the problem. :)
Your machine is blue screening, or, more accurately, hitting a system STOP condition. That's usually a driver or hardware, about half of the time the video driver is corrupt, and the other half, there are hardware issues in play. You can learn more by doing this:
Do this: Right click My Computer. Go to Properties, then Advanced, then Startup and Recovery.
Deselect the option to "automatically restart" on system failure.
When the system dies again, the traditional Windows blue screen will happen, which is essentially a memory dump. The first two lines of junk at the top should contain a STOP: error (but the screen may look a little different and provide more data and in that case the information is under Technical Information) Take note of the two sets of hex numbers following the STOP and google them. You'll either be pointed to a driver problem or a memory problem. Either way, you'll get at least an idea of what to do next.
Also, try booting the machine into safe mode (F8) after the BIOS test) and running it for a while that way. Safe mode disables many drivers and if the machine consistently operates properly in safe mode, that points to a driver issue.
posted by disclaimer at 8:33 PM on May 21, 2006
Your machine is blue screening, or, more accurately, hitting a system STOP condition. That's usually a driver or hardware, about half of the time the video driver is corrupt, and the other half, there are hardware issues in play. You can learn more by doing this:
Do this: Right click My Computer. Go to Properties, then Advanced, then Startup and Recovery.
Deselect the option to "automatically restart" on system failure.
When the system dies again, the traditional Windows blue screen will happen, which is essentially a memory dump. The first two lines of junk at the top should contain a STOP: error (but the screen may look a little different and provide more data and in that case the information is under Technical Information) Take note of the two sets of hex numbers following the STOP and google them. You'll either be pointed to a driver problem or a memory problem. Either way, you'll get at least an idea of what to do next.
Also, try booting the machine into safe mode (F8) after the BIOS test) and running it for a while that way. Safe mode disables many drivers and if the machine consistently operates properly in safe mode, that points to a driver issue.
posted by disclaimer at 8:33 PM on May 21, 2006
One way to find out if it's a heat problem is pretty straightforward: leave it off overnight, then turn it on and wait to see how long before it bags it.
posted by Steven C. Den Beste at 8:55 PM on May 21, 2006
posted by Steven C. Den Beste at 8:55 PM on May 21, 2006
I second the bad, but not visually bad capacitor. Especially if it is a known problem with the model. They don't have to leak or be grossly over sized to be bad. The whole industry has bad cap issues dating back a few years.
posted by Gungho at 7:51 AM on May 22, 2006
posted by Gungho at 7:51 AM on May 22, 2006
This thread is closed to new comments.
Try running memtest86 overnight.
posted by AndrewStephens at 6:50 PM on May 21, 2006