IntelliStation M Pro reboots randomly in XP Pro

[lparsons]

I am using an IntelliStation M Pro, type 6229 at work. Currently has 1 gb of RAM and a variety of specialized hardware for its task at work.

This sytem has a nasty habit of rebooting randomly without warning or logging anything. Fortunately it hasn't happened while I have been at the system yet, though it soon will be running overnight jobs that could potentially be severely disrupted by an unscheduled reboot.

The system is connected to a Toshiba UPS (the size of a small fridge), so I know it isn't losing power. There is another IntelliStation M Pro in the same room that generally does not exhibit this behavior and is not on a UPS at all.

A while ago I downed the system on purpose to physically examine the hardware. Everything was seated well. I then used compressed nitrogen to blow dust out of the system.

There is very little rhyme or reason to when the system reboots. It generally happens between 9pm and 8am, though that is not always true. It can happen any day of the week. I will come back in to the room where the system is and I will find it is back to the Windows XP login screen. I even set windows to not do automatic reboots and the system still reboots itself rather than crashing to a BSOD.

I just installed AVG (free version) antivirus recently. A full system scan turned up several suspect files ("Win32/Heur") that were quarantined. Otherwise nothing interesting. I also ran spybot and saw largely the same.

System is running XP Pro, SP2. Automatic updates are off, as is the firewall (we have a corporate firewall, and the specialized hardware for this system does not cooperate with a system running a firewall of its own).

Any ideas what I should look for? I have looked all over the place and have yet to find anything helpful for this.

thank you

[leoblob]

To me, random re-boots sound like a hardware issue. Just some general ideas here...

I would consider memory as a first suspect. Remove and re-seat all memory sticks. I would also run a memory diagnostic that exercises the memory for hours and hours. These diagnostic programs have been discussed here (somewhere in the off-topic area).

I would consider heat as a second suspect. I know you already blew dust out of the case, but I would carefully inspect the processor heatsink(s). On my Intellistation Z, the heatsinks have very narrow tightly spaced heat sink fins. To blow the dust out of them requires a direct shot of air (or nitrogen) aimed "backward" into the case through the heat sink fans.

While you have your compressed air/nitrogen source out, I would blow out the power supply, again by aiming "backward" into the exit fan on the back of the case.

Make sure the heatsink on the video card is clean, and if it has a fan, verify that the fan actually runs when the power is on.

I would also verify that the heatsink(s) are properly tight against the processor(s)... that the fasteners are not loose, nor have their seats pulled out of the motherboard.

[Harryc]

Power supplies in desktops have been known to cause spontaneous reboots as well. Just something to keep in mind if the tests and procedures leoblob suggested don't pan out. Also run all of the PC-Doctor tests.

[lparsons]

After we had eclipsed previous records for uptime, I came in this morning and found that the intellistation had rebooted twice this morning (twice within 30 minutes, to be more specific).

So it appears the new power supply did not solve the problem on its own - though it lead to a huge improvement. I guess it sounds like memory might be my next suspect - and of course the old RIMM's are almost impossible to come by now.

[lparsons]

I just tried a couple things, after the most recent post.

First, I opened up the case to check that everything was seated correctly, and reseated the CPU and RAM. I noticed one thing odd when I reseated the CPU - the CPU came out with the heatsink/fan without lifting the lever for the ZIF (do they still call it ZIF?) socket.
I then proceeded to separate the CPU from the heatsink, and reseat the CPU in the socket. I did not have any arctic silver (or comparable) available, so I just placed the heatsink/fan on top of the CPU and tightened everything down.

Second, I proceeded to run memtest86 and the windows memory test on the system. No faults were found in the memory by either. Memtest86 took a little over an hour (1gb of RAM).

The heatsink/fan situation leaves me with another question - are there software utilities for the following:
*cpu clock speed adjustment
*cpu temperature monitoring
*cpu fan speed adjustment

And of course I would want those to be utilities that I can run in my OS (xp pro in this case).

[virge]

. . . I then proceeded to separate the CPU from the heatsink, and reseat the CPU in the socket. I did not have any arctic silver (or comparable) available, so I just placed the heatsink/fan on top of the CPU and tightened everything down.

I did that once so just a few words of warning.... This was back when I had a 486. The computer died soon after and when I took the heatsink off I found a nice crack down the center of the ceramic backing of the CPU. And that's when I learned the importance of thermal grease.

[lparsons]

I did that once so just a few words of warning.... This was back when I had a 486. The computer died soon after and when I took the heatsink off I found a nice crack down the center of the ceramic backing of the CPU. And that's when I learned the importance of thermal grease.

I honestly couldn't decide how much to be concerned about thermal grease anymore. The intellistation in question has a pentium4 CPU (as opposed to our other intellistation which has a xeon). Earlier in the week I pulled a pentium4 from a dell (optiplex) workstation, and found that dell apparently does not use thermal grease for their pentium4 systems. Not that I am a big fan of their work, but we haven't seen any dells go down to heat (yet).

I'll probably pick up some thermal grease when I get a chance, just to be safe. Though the way that these sockets work in regards to heatsinks and fans leaves a question - how is one supposed to remove the fan if it is held into the CPU by thermal grease? IBM of course is fond of the heat pads on the laptops, and whatever they use on the intellistations seems to have similar adhesive properties.

This question comes up as a follow-up to my observation that pulling the heatsink/fan caused the CPU to come right out of the socket, which I doubt is very good for either the CPU or the socket.

I know I had a tool for monitoring system temp on my thinkpad. I'll have to look and see if something similar exists for the intellistations, I think that will help address this concern.

[leoblob]

I agree that it's not a good idea to run the computer without a new application of thermal grease. I think the reason the heatsink pulled the processor out of the socket is because old thermal pads can get somewhat adhesive after a while so it's difficult to separate the processor from the heatsink.

At your earliest convenience, I would carefully clean off the heatsink and processor mounting surfaces with alcohol and then apply a thin layer of really good grease.

I'm not sure why the new power supply made things better but did not completely fix the problem. In the meantime, however, I would definitely re-install the heatsink with new grease, to avoid creating a new problem.