Mini PC Crashes

OK, we need to wait for a couple more restarts then. Photograph whatever is on the screen if it tries to restart and then run the Sysnative collection app again after a couple of these restarts.

Please don't do anything that we don't ask you to do until the issue is fixed.
 
ok, so the computer has crashed several times. I have collected the data and attached. When it reboots, it just stops and restarts. There is no opportunity to take a picture. That is why I thought the issue was hardware, so I replaced the power supply and the memory. Not much else to replace on a mini PC except the m.2 drive.
 

Attachments

Can you just confirm that the 'Automatically restart' box mentioned in my last post is NOT checked? If it still restarts then this isn't a Windows system failure causing these restarts.

You said earlier that the temperature sometimes hit 100°C, what temperature were you measuring there; the CPU, the mainboard, etc.? Overheating is a common cause of sudden shutdowns, it's a protection mechanism. They don't usually result in restarts however. I'd suggest you get a decent temperature measuring tool (HWMonitor free is good) and post two screenshots of the CPU temps (and any others that look significant) at two different times...
  1. Allow the computer to go fully idle, have nothing running (except for normal background tasks). Post a screenshot of the temperatures at idle.
  2. Start as much demanding work as you can - we want to push the computer as hard as possible so that components get as hot as they are ever likely to. Post a screenshot of the temperatures at full load.
I think the most logical next step is to check to see whether there are any rogue drivers that might be causing these restarts, so I'd like you to enable Driver Verifier using the instructions at Driver Verifier Instructions - BSOD - Windows 11, 10, 8(.1), 7 and Vista. Have a read through them first so that you get the full picture and them enable Driver Verifier using only the options in that link.

We expect BSODs with Driver Verfier enabled, that's what it's designed to do, so be sure to save all the dumps. In your case I would leave Driver Verifier enabled for 48 hours, or until you've had at least two BSODs of the special types described in the above link. For now, ignore any other restarts that may happen.
 
Hi Ubuysa,
yes, I can confirm it is checked. See screen shots. I thought I had posted the other day with screenshots, but I don't see it. I have core temp set to log upon login. I ran heavy load to keep the machine at about 85% for an hour. It never went over 90. Still didn't crash. I ran drive verifier. I am not sure what I should see, but it is running.
 

Attachments

OK, it's probably not overheating then. If Driver Verifier finds any bad drivers it will BSOD. However, Driver Verifier can only check drivers as they are loaded, so it's important to get every third-party driver loaded at some point. That means you8i need to use every third-party app, every device, every feature to be sure that all drivers are loaded at some point. If you do that and there have been no BSODs after 48 hours you can disable Driver Verifier and we can cross a bad driver off the list too.

One other thing; I see you installed new RAM? Are you 100% certain that the new RAM is fully compatible with that mini C? All RAM is not the same, you need to verify that your new RAM is fully compatible. TBH, unless you know it to be faulty, it might be wise to replace the original RAM.
 
Hi Ubuysa, Yes, I bought Kingston RAM that was recommended by Gigabyte. I ran the Microsoft memory diagnostic and memtest on it. Both came up good. I booted the machine to fedora using a thumb drive and it has been happy with that all day. I wanted to see if it would crash with another operating system to see if the issue was hardware. Since it is happy, I think we can rule out hardware. Of course, the one piece of hardware not being tested is the M.2 drive where the OCS is installed. Samsung Magician thinks it is working perfectly, but, maybe it is faulty? Could that be the issue? In which case I guess I would get a new drive and rebuild the machine. I can restore from backup, so it might not be horrible.
 
I agree then that the base hardware is fine. That was a very good test to run Linux.

I see no indications that the M.2 drive may be at fault and I don't suggest replacing it (yet).

The one dump in that last upload showed that a networking operation was in progress, and a USB3 device too it seems (although on-board network adapters can appear as USB devices).

I think the best thing to do now is to enable Driver Verifier to see whether it can catch a rogue third-party driver. The instructions on running Driver Verifier are at Driver Verifier Instructions - BSOD - Windows 11, 10, 8(.1), 7 and Vista. Have a read through first to be sure you understand what you need to do - it's not as complex as it may seem.

Driver Verifier can only test drivers when they are loaded - so it's essential to get every third-party driver you have loaded at some point. That means using every third-party app, every device, every feature that you have, in order to get all drivers loaded at some point. Driver Verifier will generate a BSOPD if it finds a a flaky driver, look out for the bugcheck codes listed in the above. Keep ALL dumps! After 48 hours you can disable Driver Verifier if it's found no rogue drivers.

See what that finds....
 
OK, so I have good? news. The computer crashed with Fedora. It might have run out of disk space because I was logging, but it crashed. Drive Verifier has been enabled for a while and no BSOD yet. But, the computer doesn't stay up for 48 hours no matter what I run as an OS. So, to reiterate - I have a new power supply, it doesn't appear to be overheating, it has new memory, and it crashes on Fedora live and Windows. Sounds like hardware, but the only thing it could be is the processor. Does that happen?
 
Yes, it can be the CPU. The symptoms certainly exist. The good news is it can be replaced on a mini. If you feel up to it, you could do it yourself. If not, make sure the repair person has worked on mini's before. Still exhaust all of Ubuysa suggestions before doing that.
 
I am running Fedora again. I created a partition on the SD card for pmlogger to record. I will get data every second or so, so I can see when it goes down and what it was doing. If not, let's try some more Windows investigation.
 
Ah, that does sound like a hardware issue then. You might want to move over to the hardware forum then. Post a link back to here so that people can see what you've already done.
 

Has Sysnative Forums helped you? Please consider donating to help us support the site!

Back
Top