Loss of response to keyboard and mouse

[milleron]

My new A8N-SLI system recently started to cease responding to input from keyboard and mouse after it's been on overnight. This happens whether it's idle or running a task like Prime 95. The next morning, the screen will be black, and it's impossible to "wake" the computer with either keyboard or mouse.
There is no BSOD.
The computer is not in Standby mode because the fans continue to run.
A possible CLUE is that hitting the power button turns off the machine immediately whereas it must be depressed for more than four seconds when it's running normally.

The Corsair RAM passes 10+ hours of Prime 95 Torture Test.
The voltages on all rails are nearly perfect per MBM5, but a little low per Asus Probe -- i.e, varying from spec by 4-5% rather than the 3% guaranteed by the Antec TPII 500.
I've disabled CnQ without benefit.

Can anyone give me a good guess where to start the troubleshooting?

Ron

 

[porcorosso]

Your clue about the system shutting down immediately when the power button is touched is significant, I think. AFAIK that shouldn't happen if an ACPI-compliant operating system is still in control of the system -- meaning, I suppose, that the system isn't really in any kind of standby but is crashed. Is this system running Win2K or WinXP? Something else? If Windows, what HAL is indicated in DM? Also, so you see anything in the Event Viewer?

Are the mouse and keyboard using USB or older style connectors? IF USB are any special software loaded for them, or are they running as HID devices?

Besides CnQ are there any other non-OS-integrated third party utilities that might be affecting power management?

I'm going to watch this thread closely because I've seen similar behavior on a couple of systems before, but never with any kind of regularity. Sounds like it happens every time this one is left on overnight?

 

[Jiggz]

Just because the fans are running it doesn't mean the system is not in standby since there is an option in the bios to either have the fans running or not. The symptoms indicates the system is in stby but will not wake up from stby. So you need to troubleshoot by purposely placing the system in stby and then see if you can wake it up from stby with the kb or mouse.

 

[porcorosso]

Good point about fans running in some levels of standby, Jiggz. But what do you make of the fact that the system shuts down immediately if the power button is touched? Should that happen if the system is in standby? Haven't really noticed or tested this before, but I assumed that you would still have to hold the switch in position to force a shutdown.

 

[furballi]

Back off on the OVERCLOCK by at least 50MHz CPU core speed. If problem persists then do another 50MHz.

Also uncheck USB power saving options in DEVICE MANAGER and any standby mode in BIOS.

 

[milleron]

Originally posted by: porcorosso
Your clue about the system shutting down immediately when the power button is touched is significant, I think. AFAIK that shouldn't happen if an ACPI-compliant operating system is still in control of the system -- meaning, I suppose, that the system isn't really in any kind of standby but is crashed. Is this system running Win2K or WinXP? Something else? If Windows, what HAL is indicated in DM? Also, so you see anything in the Event Viewer?

Are the mouse and keyboard using USB or older style connectors? IF USB are any special software loaded for them, or are they running as HID devices?

Besides CnQ are there any other non-OS-integrated third party utilities that might be affecting power management?

I'm going to watch this thread closely because I've seen similar behavior on a couple of systems before, but never with any kind of regularity. Sounds like it happens every time this one is left on overnight?

OS=XP Pro, SP2, all critical updates installed
HAL=ACPI
Event Viewer gives no clues (also, it doesn't indicate that the system entered Standby)
Standby in BIOS = S1 & S3
XP Power Management was "Minimal" when CnQ enabled; now "Desktop" with Standby set to "Never." Neither setting makes any difference.
Mouse and keyboard are PS2 connected through IOGear KVM (worked with KVM for 8 mo)
I don't know of any other third-party power management applet.
Regarding regularity, I'd say it's a 90% thing. I have occasionally found it responsive in the AM since this started happening a few weeks ago.

I was thinking this is a hardware problem, but your point about the computer not being under the control of an ACPI-compliant OS -- i.e., CRASHED -- is very cogent. If that happened, I'd expect to come back to a BSOD, though, wouldn't I? Is it possible that such a BSOD is occurring and that the monitor is blank simply because it was already in sleep mode when the crash occurred?? (Monitor is a Samsung SyncMaster 191T.) I'll see what Dr Watson can tell me.

Thanks very much for your interest and help.

Ron

 

[milleron]

Originally posted by: Jiggz
Just because the fans are running it doesn't mean the system is not in standby since there is an option in the bios to either have the fans running or not. The symptoms indicates the system is in stby but will not wake up from stby. So you need to troubleshoot by purposely placing the system in stby and then see if you can wake it up from stby with the kb or mouse.

I can't find that setting in BIOS (1009). When I manually put it in standby, it wakes up just fine, but I haven't set it to standby manually and then waited overnite yet. That's an easy troubleshooting step. I'll do it and report back.

Thanks!

Ron

 

[milleron]

Originally posted by: furballi
Back off on the OVERCLOCK by at least 50MHz CPU core speed. If problem persists then do another 50MHz.

NOT overclocked at all. Memory timings are at default except for 1T command rate. All settings the same since system was built in June 2005.

Also uncheck USB power saving options in DEVICE MANAGER and any standby mode in BIOS.

I've not noticed these Device-Manager options. I'll look for them and uncheck them. I did get a new USB scanner recently, and its cable is left connected, but this happens even when it's powerd off. I'll disconnect that cable just for the heck of it. I'll set the BIOS to disable Standby.

Thanks!

Ron

 

[Jiggz]

Originally posted by: milleron

Originally posted by: Jiggz
Just because the fans are running it doesn't mean the system is not in standby since there is an option in the bios to either have the fans running or not. The symptoms indicates the system is in stby but will not wake up from stby. So you need to troubleshoot by purposely placing the system in stby and then see if you can wake it up from stby with the kb or mouse.

I can't find that setting in BIOS (1009). When I manually put it in standby, it wakes up just fine, but I haven't set it to standby manually and then waited overnite yet. That's an easy troubleshooting step. I'll do it and report back.

Thanks!

Ron

This options depends on the type of mobo but generally there are options where in you can power up the system using the mouse and kb. There is also an option about resuming using USB devices. As for the power button, there are options wherein by pushing the power button you can instantly turn off the system or using a 4 sec delay. And then there is the ACBACK Funtion option of making usre the system is in Full On state when power is resume or stay in Full Off when power is resume. This options are all under the Pwer Management Setup in the Bios.

 

[milleron]

Dr Watson logs are practically indecipherable, but I note an entry for 10PM last night when I was away from the computer. A few hours after that entry, its screen was black and unresponsive. That log entry contained references to several system files. Almost all the files other than system files were from Norton AV, Ghost, or Norton common files. After rebooting the machine, I noticed Norton AV and Ghost tray icons closing with messages about sending reports to MS. This was followed by a similar notice from the Windows Defender EXE file.
Well, all those Norton products have been installed from the get-go, but Windows Defender has been here only since the upgrade from MS Antispyware, perhaps around the time my problems began. I uninstalled Defender last night, and the computer responded to input and awoke normally this AM. That's not unheard of, but it's unusual.

Since I may be on to something here, I'm suspending other troubleshooting for a day or so to see if the computer still crashes overnight following removal of Defender.

 

[w00t]

To get accurate Psu Rails you would have to use a Digital Multimeter.

 

[milleron]

Originally posted by: w00t
To get accurate Psu Rails you would have to use a Digital Multimeter.

Thanks very much for the suggestion. Unfortunately, I'm a health-care professional rather than an engineer. I might be able to figure out the multimeter, but I'd still not know where to measure the voltages. I imagine that explaining that process is beyond the scope of this forum.

 

[furballi]

Measure at the MOLEX plug (go to the HDD). Black-Yellow is 12VDC. Black-Red is 5VDC.

 

[milleron]

Uninstalling MS Windows Defender didn't stop the problem. This evening I came back to the computer. The screen was blank (monitor asleep), and it did wake up normally with keyboard input, but it immediately blue-screened with the dreaded "IRQL_NOT_LESS_OR_EQUAL." It seems like that's a video problem as often as not, so this evening I'm upgrading the reliable Catalyst 5.9 to whatever ATI has, presumably 6.3.

 

[porcorosso]

Sorry I didn't respond before now. I didn't see an e-mail notification.

Your answers have pretty much negated the possibility of most of my suspicions being the culprit. With the additional information that you and others have provided I'm thinking you've got a software / driver problem. The first thing I'd do (personal bias being a strong factor here) would be to get rid of anything Symantec / Norton and try the system running without them (and without Internet connection, too, of course).

If someone hasn't suggested it already, you might try resetting the BIOS to its default settings and trying the system for a couple of days running those. Maybe even with standby and hibernation disabled in Windows, and with power saving settings for all devices that support it disabled, too. I would be curious to see if this condition is brought about by BIOS settings , OS settings, or both, or neither. It would take a few days to find out, I guess.

I'm afraid I really have little of value to contribute but will continue to watch the thread in hopes that you find a solution.

 

[milleron]

I thoroughly agree with you. I share your bias against Symantec. The misery I experienced with this machie and Ghost 9 filled a multipage thread here last summer, but I thought that Ghost 10 was working well.

My next troubleshooting step is to remove all the Symantec stuff. I'm going to do that on April 14, and I'll post the results here on April 16 (it really takes 48 hours to tell if any change affects this problem).

Thanks for staying with me on this one. My computer is very reliable. I've never lost any data with these crashes, but I want to make it 100% reliable like it has been for most of its existence.

Ron

 

[porcorosso]

I will keep tabs on this thread. I'm very interested in seeing how this turns out for you.

Ross

 

[milleron]

Ross,

Last-second course correction . . . .

I have Corsair XMS RAM. Both DIMMs failed after I built my rig last summer, and Corsair replaced them both. My problems then were a little different inasmuch as I could not make the Prime95 Torture Test run for more than a few minutes without a failure. Now it will never error out. However, some of the actual machine crashes I saw were the classical "IRQL_NOT_LESS_OR_EQUAL" like I've seen twice here in the last week, even after upgrading the Cat driver. This makes my feeble brain think that perhaps there could be a RAM problem again. Because Prime95 runs perfectly, I know the DIMMs are intrinsically healthy, but I'm wondering if a tad more Vdimm could help. The RAM is already running at 2.8v which is above the default setting, but the techs at the Corsair forum often recommend 2.9 for troubleshooting. I'll run it like that for two days. If no joy, I'll then trash Norton.

Ron

 

[porcorosso]

Hi, Ron.

Sounds like a plan. No hardware guru is I, but I have definitely seen replacement RAM fix issues in WinXP -- despite the fact that the original RAM showed no errors under Prime95 or Memtest. WinXP seems to be more of a torture test for many systems than any diagnostics software out there.

Good luck!

Ross

 

[imported_Imp]

Since everyone else has offered a pretty complicated explanation, I'll throw in my half-cent. Sometimes (ok, a lot of the time), it's probably just a simple defective part causing all the problems. In this case, sounds like either your mouse or keyboard is dying and screwing everything else up.

This happened to me on my old system, my mouse was dying, but it worked properly the other 50% of the time. This POS of a mouse started skipping when "working", and half the time it just suddenly died while the application kept on running. Note that it also took the keyboard along with it, and managed to lock the entire system up too. The second I replaced the mouse, no skipping, no crashing, perfectly stable again. I spent months checking for overheating or other causes, turned out it was just the mouse. More recently, my craptastic mobo was part of a crap batch with defective SATA controllers which screwed me over for 2 months. Put in IDE HDD, gone.

So, my suggestion is that you at least try to switch out either your keyboard or mouse before doing anything drastic. Sounds like your system is going in sleep/standby, but can't receive any input to wake up.

 

[milleron]

I went through two days of hardware troubleshooting by booting to a UBCD4Win that runs an XP OS from the CD and a RAM drive. I set it to run one of its self-contained utilities, MemTest (different from MemTest86) and let it go for 48 hours while actually using the computer for checking email and Web surfing. The computer was rock stable, and MemTest returned zero errors. I felt that this gave a clean bill of health to motherboard, RAM, PSU, video card, and peripherals.

Because I had seen Norton Ghost's (ver 10) tray applet give error messages, followed by a cascade of other apps closing (Intellitype, Zone Alarm, ccApp), and because the preceding version of Ghost, 9, NEVER worked on my computer, I selectively uninstalled it while leaving NAV and NU installed and running. It's a little early to say that the problem's cured, but the computer's run overnight twice now, and appears to be perfectly stable at 52 hours or run time.

This may not be the fault of Ghost 10, itself. It could be a conflict between Ghost 10 and Acronis True Image 8 which is also installed. It could be a conflict between Ghost 10's tray applet and a tray utility. And, of course, the computer might crash tonight, sending me back to the drawing board.

For now, I'll let it run this way for a few weeks. If everything goes perfectly, the final troubleshooting step would be to reinstall Ghost to see if the problem can be reproduced.

Ron

 

[milleron]

For anyone still subscribed to this thread, I wanted to post the final solution.

After a very long series of troubleshooting steps, it was obvious that the computer was, indeed, crashing. The error messages suggested RAM problems. After finding errors in both Memtest-86 and Prime95 Torture Test, Corsair made me test the DIMMs individually. Both checked out perfectly, leading to the conclusion that there was a motherboard problem. THEN, after reinstalling both together in dual-channel mode, Memtest-86 and Prime95 each ran without errors for >12 hours each, AND Windows became perfectly stable again.

Conclusion: the weird problems were caused by electrical-connection issues that were corrected by reseating the DIMMs (as part of the individual testing mandated by Corsair). I always reseat PCI cards, cables, and power connectors early on in any troubleshooting situation, but I never thought to do that with DIMMs because they are prevented by their built-in locking mechanism from coming loose. Nevertheless, gradual degradation in an electrical connection in the DIMM socket seems to be what was causing this problem.

Ron