Stability issues

[Yannai]

My server restarts itself regularly, but at odd intervals (it can go weeks without restarting and then go multiple times in a day). I don't think its a virus (I now use AVG) but might be some other malware. I am running it on a cheap refurbished emachines so it could be a power supply, RAM, or other problem. The only other thing I can think of is that my system hard disk runs hot (70 degrees C).

Anyone have any advice on how to diagnose this problem.

[CollinR]

emachine model number.

[Yannai]

It is a C3070

Sempron with an nvidea MB

[MeInMaui]

You can try checking your system logs under

Control Panel >> Administrative Tools >> Event Viewer

Look for errors and see what the log says. It might help.

Aloha,
Mike

[Yannai]

My Server crapped out and wouldn't even boot past the drive selection screen a few nights ago. turns our one of the Ram sticks was bad. Seems to be running stable now, so hopefully that was the problem.

[m1abrams]

When debugging stability issues where the entire OS blue screens (2000/XP) which is what is happening if your machine restarts (default behavior on XP is to restart if it blue screens). I check these hardware parts in order:
1. Visually inspect the motherboard. You want to look for burn marks or sticky yellowish brown stuff on the board. Pay close attention to the capacitors, many motherboards have failed due to cheap caps. You can tell if they have failed because either the top is bulging or leaking fluid, or I have seen them blow out the bottom and you will see fluid or burnt black stuff out of the bottom.
2. Harddrive - run the drive manufactories test program, Ultimate Boot CD has all the popular ones on a single iso image. If you own a copy run Spinrite
3. Check the memory using memtest, again Ultimate Boot CD has this, or many linux distros like ubuntu have this as a boot option. Run this test over night.
4. Check the power supply, this one is a bit tougher to test if you do not have a load testor. You can not just connect a DMM to it and tell, you need to apply a load to the power supply. Sometimes you can tell if you look at the voltage readings from the motherboard but those tend to be off a bit so you would have to know if the voltage has varied from the norm. Which would require you to know what is norm for your motherboards readings. If it comes to this I usually see if I can get a Linux live CD and run load tool to run stable for a day or so. If it does then I would reinstall the OS. If it does not run stable then if the memory, drive check out and the motherboard "looks" ok then I would just buy a new PSU.

Note: because the memory test and harddrive test rely on the motherboard and PSU functioning correctly they may fail and it not really be the fault of that part. Which is why I say visually inspect motherboard first. When doing the memtest and harddrive test disconnect unneeded device from the PSU. Like harddrives you are not testing (all in case of the memtest), CDROMs not being used, usb devices cameras etc. Unplug the onboard tuner cards. By reducing the load you can help reduce the influence of a flaky PSU during the test.