0

I have a machine that has been running fine for the last few years as a home server and now it has problems with freezing and restarting continuously.

I want to figure out what the problem is with the machine.

The specs are:

  • AMD Sempron CPU
  • 2GB RAM
  • Windows 2000 Server
  • 20GB HDD (C:)
  • 2x 250GB HDDs in Raid 1 configuration (D:)

It's hard to diagnose the problem from within Windows because the machine could restart of freeze at any time.

I've run Memtest86 from a bootable CD and that worked fine - no errors after 13 hours and 11 passes, but trying to boot from an Ubuntu live CD didn't work when I tried it. It halted with a text dump screen of some sort displaying the IP and some memory addresses (can't quite remember exactly what was one there).

Is there some program I can run from a bootable CD that will diagnose the problem and tell me which piece of hardware is faulty?

What can I do to quickly diagnose the problem?

Update

I finally repaired the machine. Although the PSU had tested ok with the PSU tester, a few days later when swapping out the motherboard it blew up with a pop and a cloud of sparks and smoke.

I replaced the blown PSU with an Antec EarthWatts 500W PSU and was able to get the system functioning.

The graphics card was also a bit dodgy so I ordered in a second hand replacement.

So in summary, I replaced the following parts:

  • PSU
  • Motherboard
  • AGP graphics card

and was able to get the machine running. The system test software that I used was unable to detect the faulty hardware. Luckily before hooking up the SATA drives I did a backup because setting up RAID on the new motherboard wrecked the data. I had to fiddle around with drivers and Windows to get everything working correctly.

1
  • Memtest needs to run overnight to stress test the memory.
    – Moab
    Aug 8, 2010 at 15:20

2 Answers 2

1

There are plenty of PC diagnostic programs out there, but they all suffer from the same limitation: in order to run the program, most of the important components of a PC have to already be working properly. It's kind of a catch-22. A truly useful diagnostic for a PC would consist of hardware that is separate from and monitors the main PC hardware.

If, as you say, the computer has been working fine for a while, and nothing on the computer has changed recently, then it sounds like the power supply has gone bad. A bad PSU can cause all kinds of weird, intermittent problems. Normally for random lockups I would suggest display problems, but if this is a server, that means you don't use it directly in normal use - is that correct?

PC PSUs go bad or flaky all the time. Depending on where you live, the AC coming into the building may not be too reliable and typical PSUs are sensitive to major fluctuations. A lightning strike nearby could certainly do the trick. The protection offered by a typical power bar is not worth much.

You can buy a power supply tester (about $30), swap the PSU with a spare if you happen to have access to one, or buy another PSU ($100) and swap it.

2
  • Yes, it's being used as a server and usually there is no monitor plugged in because on the rare occasions I log into it I use Remote Desktop. I'll have a look for a power supply tester. I didn't realise such things existed. Aug 9, 2010 at 11:59
  • I have now tested the PSU and the tester gives it the all clear. I will replace the video card next with a cheap one second hand one and see if that makes a difference. Aug 11, 2010 at 13:22
1

strip the hardware down to bare essentials - disconnect the raid drives, take out the nic, disable/remove the audio card, reduce ram to minimum (if there's two sticks, take one out), unplug all peripheral devices. see if you can boot from a very recent linux live cd (ie ubuntu 10.04). use the boot option on the cd to launch in vga mode. let it run intil you're satified its stable, then start adding back one device at a time. Since its a server and you said the Linux error mentions the IP address, I'd start with the NIC.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .