Interesting computer deaths after shutdown

[lakedude]

We have racks of computers at work that will run forever, so long as you don't turn them off. Unfortunately I have the assigned task to reboot them weekly and very frequently one will not come back on.

This is very curious because the failure mode is always the same. The computer runs fine, reboot and then the computer does not boot or have any video. It is too noisy to hear any beep codes if there are any.

If you put the hard drive from a bad computer into a good chassis the good one will work fine, eliminating software and the hard drive as possible cause.

Since these are not my computers I can't really play with them to determine much about the hardware.

A failed computer outputs no video, none. No error messages, no boot sequence, nothing. The power led will come on and the drive light will flash a few times and then nothing.

The computers are monitored remotely and the bad ones never come back up, it isn't just a video issue.

I'm very curious as to what could cause a problem like this. I wonder if the computers are failing because they are turned off, or if they have already failed and the failure is not noticed until the reboot.

What do you think?

 

[John Connor]

Bad PSU, GPU or MOBO?

 

[Ketchup]

Age, dusty environment, heat, bad memory, lots of things. Maybe you need to discuss lengthening the reboot frequency.

 

[lakedude]

I peeked inside one of these. They are kept in a proper server room that is clean and temperature controlled. The inside of the computer is pristine, like new. The cases are excellent, they have air filters on the intakes. The MOBO is some kind of dual processor Intel board.

I powered a bad one up in a quiet location, no beep codes. Fans come on, drive indicators flash, power led stays on but no boot or video.

 

[lakedude]

Now we are getting somewhere, MOBO is Tyan S5360. Similar to this: Tyan Thunder i7520R (S5360-1U) Server Board Intel Socket 604 800MHz FSB 16GB DDR SDRAM (Refurbished) Mfr P/N S5360-1U

 

[lakedude]

SOLVED! I booted with only one stick of memory and the system lives. Looks like a memory or memory slot issue...

 

[lakedude]

The system boots with 1, 2, or 3 sticks but it does not boot with the 4th stick in place.

 

[lakedude]

Fixed! System is running after replacing the 4th memory module. +1 Ketchup!

 

[lakedude]

The memory is Kingston KVR333S8R25/512 and it is either semi incompatible or it is garbage. These things go out all the time. Is this a known issue?

 

[Ketchup]

Most memory has a lifetime warranty. And Tyan is known for goot server-grade products. It's going to take more than one stick of memory to find a pattern, but good job with the first fix.

 

[lakedude]

System 2 also fixed, had 2 bad sticks, both Kingston KVR333S8R25/512...

Normally I'd just be swapping these systems out for rebuilt spares but we have a software upgrade due and we have no spares so time is of the essence.

This happens very frequently, systems are failing before we can get rebuilt units back. Technically I'm not the one to be fixing these systems but someone needed to.

Thanks for your help Ketchup!