Weird hang after various lengths of time

[prouton]

I have a client who is experiencing the strangest "hang" problem I've ever run across. When it hangs, the mouse pointer still moves, the Start menu still responds, system tray icons will pop up their menus, and ctrl-alt-delete brings up the Windows Security dialog. But when the Task Manager button is clicked, nothing happens except the security dialog goes away. Clicking on any menu item closes the menu, but nothing starts. The hard drive is not being accessed during this "hang" period, and since I can't get Task Manager to start I can't tell if the CPU is doing anything. The only way to powerdown the machine is to hold the front panel power switch down for several seconds until the machine does an abrupt power off. Then, the next time the machine starts, it takes several minutes (four or five times longer than normal) for the machine to reboot into Windows. The machine does not indicate it is running chkdsk/scandisk, but I can't tell what it is running that takes so long.

Operating system is WinXP sp1. Various applications used through the day are AOL 9, MS Office 97 sp2a, Roxio, QuickBooks. Computer is an E-Machines Celeron (2.6GHz, 512meg RAM) roughly 6-9 months old. Both AdAware 1.06 and Spybot 1.3 think the machine is whistle clean, as does AVG 7.

The time since boot it takes for the problem to occur varies from thirty minutes to several days. Cold booting the machine is the surest way to correct the behaviour. The computer developed this problem over multiple weeks, with no hardware updates in that time, and few if any software updates.

I'm stumped as to how to proceed.

[104456]

Sounds like its having a harddrive access problem so Id suggest testing the drive for errors using the manufaturers tool [links here ]

[jmwills]

I would have you run a file integrity check from the run command "sfcscannow". Have you Windows disk ready as this will repair all your systems files.

[prouton]

Thank you both. I'll try the suggestions and let you know.

[prouton]

Well, I ran the Hitachi Deskstar (drive model is 7K250) disk check utility and it gave the drive a clean bill-of-health. But what a pain running that on a system with no floppy, and only the WinXP built-in CD writing software. Hitachi does provide an ISO image, but of course WinXP doesn't know what to do with those. I eventually found/used ISO Recorder Powertoy (this is the version for SP2 systems) and away I went.

As for sfc.exe /scannow, I can report something positive for E-Machines of recent vintage: they include the I386 folder inside the Windows folder, and have the registry setup correctly to find it. What I don't know is if I missed some results information because of how I started it (because the program closed after it finsihed without giving me any feedback, and there was no log file I could find). Thinking this was a DOS level program, I started an instance of command.com from Start/Run, and then typed the sfc command there. It opened a true window in which it report the progress of the scan, but as I mentioned, it simply closed after reaching 100%.

However, that doesn't seem to have solved the problem as the computer later in the evening exhibited the problem behaviour once more. Well, twice more. The only other connection I can sort of make is that AOL 9 was either being asked to do something, or was being shut down at the time of many of the hiccups. The only time I've witnessed the problem when AOL wasn't involved was shortly after a reboot, with a newly installed copy of Adobe Acrobat Reader 7 open, and things hung. Task Manager could not be started, although after waiting for several minutes I tried to start it again and things resumed (and Acrobat was able to repaint its window).

I've actually started the process of weening the client off of AOL by setting up Netscape Communicator to access her AOL mailbox, but I can't believe she's alone in having this problem if it's caused by AOL.

Obviously I can't tell if the CPU is being maxed out (and giving the symptom of hanging) or if for some reason control messages are being lost (i.e., "start program x" getting dropped into the bit bucket). All I do know is that the harddrive is not being touched enough to make the access light even blink, and most of the time the system never recovers.

Any more thoughts or suggestions (oh oh pleeeeeese!).