Page 1 of 1

Weird lockup problem

Posted: 2006-12-18 09:27am
by Darth Wong
Has anyone ever seen this before? I had a machine that ran for years without a hitch. One day it suddenly started locking up every 5 minutes. No change to the OS, no change to the hardware.

I noticed that the CPU was running rather hot, so I took off the heatsink/fan, cleaned off the dried-up thermal paste, re-applied some fresh thermal goo and re-seated it. CPU temp dropped 20 degrees C, so I figured that would solve the problem. Booted it up in a different location and it ran fine, so I put it back in its original spot. 5 minutes later, lockup.

So now I wondered if there was a problem with the case or the power supply, since it was using an old oddball MicroATX case. So I transplanted the whole MB and all hardware into a new case, figuring that might solve it. Booted it up in a different location and it ran fine, so I put it back in its original spot. 5 minutes later, lockup.

Now I wondered if the motherboard was failing, so I booted it up in a different location and ran Memtest86 on it for 3 hours. Solid as a rock. So I put it back in its original spot. 5 minutes later, lockup.

At this point I wondered if the hard drive had become corrupt somehow, so I reinstalled the OS. Booted it up in a different location and it ran fine, so I put it back in its original spot. 5 minutes later, lockup.

Now I figured maybe there was something about the way it was hooked up in its normal location, so I tried some trial-and-error and discovered that it only locks up when the serial cable is hooked up to its UPS. When I was booting it up to test it on my desk, I wasn't hooking up the UPS cable. That's why it was rock-solid on my desk but locking up every 5 minutes in its original spot.

Has anyone ever heard of something like this?

Posted: 2006-12-18 10:18am
by Sikon
One web page seems potentially relevant, mentioning a situation where "if you tried a straight-through cable, opening a serial link to the UPS would be interpreted as a shutdown command." A link is here.

Posted: 2006-12-18 11:03am
by Darth Wong
Thanks for trying, but I've had that same cable for years and it never did this before. It literally just started happening on the weekend.

Posted: 2006-12-18 11:05am
by Faram
I have seen a similar problem once.

Then it was a HD that was the cause of all the troubbles, might be worth checking out.

As in two days later the HD gave a screetcing noice and died.

Posted: 2006-12-18 11:10am
by Tolya
Its probably a stupid idea, but have you considered a failing UPS? It could be just one port - the one that the cable is hooked up to.

Im no electrican, but I would check the stability of voltage coming out of the UPS. Most of the hard lock-ups that I experienced with my machine was because of the dying PSU.

Also, what kind of lockup is this? A solid-rock lock-up or does it pop some error message?

Posted: 2006-12-18 11:11am
by Darth Wong
Tolya wrote:Its probably a stupid idea, but have you considered a failing UPS? It could be just one port - the one that the cable is hooked up to.

Im no electrican, but I would check the stability of voltage coming out of the UPS. Most of the hard lock-ups that I experienced with my machine was because of the dying PSU.

Also, what kind of lockup is this? A solid-rock lock-up or does it pop some error message?
Total rock-solid lockup. Nothing in any kind of logs, no error message, nothing. The machine just transforms itself into a brick until I reset it.

Posted: 2006-12-18 11:27am
by Tolya
Then my guess its something with the power supply. I had the same kind of lock ups when my old Modecom PSU started to die two years ago. And it all started suddenly - and after a week it was dead. Back then I thought it was the mobo.

Check the power supply, if the voltage levels are unstable (unfortunately Im not experienced in this field, my dad has more experience in this but he is not around) then you have your winner.

Of course its a wild guess on my part, but if something is going wrong its the power supply. Maybe try plugging in another machine to the same spot and port as the problematic one and see what's happening.

Posted: 2006-12-18 11:35am
by Arrow
If you haven't changed the OS (I'm guessing this is Linux), then I'd say the serial port is freaking out. Have you tried hooking the UPS to another computer? If that computer crashes, then the UPS's port is bad, otherwise it's the MB's port. I've seen serial ports cause other weird, nigh-unexplainable problems in the past, so it wouldn't surprise me if this one of them.

And if this is a Window's box, there was a new batch of updates last week, and if your machine is auto-updating, one of those updates could be the source of your problem.