My system seems to crash from time to time. I still don't know what causes it.

If I leave it untouched for a few hours, sometimes, it crashes.

To resume, I have to force a reboot by unplugging the power cable (not even pressing the power button for N seconds seems to work).

Then, it seems to work just fine (after displaying some error messages about lost or orphaned inodes at boot). Until, one day, it happens again. When? I never know. It seems to follow some strange and unpredictable pattern.

Where should I start investigating?

  • Shdwdrgn@mander.xyz
    ·
    27 days ago

    Certainly seems like a hardware issue, but there's no easy answer to this one. It could be your power supply, motherboard, CPU, memory, even the video card. The power-button issue makes me lean towards power supply or motherboard though (assuming you've verified the power button works after a fresh boot).

    If you have other parts on hand (even from another running system) you could swap components until you identify the culprit. If you find it's your power supply, make sure you replace it with a decent quality one, NOT one of those $25 units you find everywhere, or you'll have even more problems followed by a rapid failure in another year.

      • Shdwdrgn@mander.xyz
        ·
        27 days ago

        Then it could still be the power supply or motherboard. It takes a lot to override the hardware power switch and the power supply itself is usually one of the biggest culprits to random lockups. Beyond that I can't offer much, I don't know of any way to test the components unless you could afford thousands of dollars for specialized equipment. I've always had multiple machines available to swap parts so I don't have a different strategy for troubleshooting.

        • Ninguém@lemmy.pt
          ·
          26 days ago

          Thanks.

          I thought there could be some place in logs the system could write error messages to. Even hardware derived.

          I've been going through /var/log/* and found thousand of scary messages, but can't really make sense of any of those.

          • Shdwdrgn@mander.xyz
            ·
            26 days ago

            Something like a hard drive could generate recognizable errors in the logs, but once you get out into the motherboard or power supply, there's simply not enough monitoring available to detect problems. You might be able to record the voltage output from the power supply rails, depending on your hardware, but that still might not tell you anything.

  • Ninguém@lemmy.pt
    ·
    25 days ago

    Thank you all who contributed.

    I was hoping for some log that could maybe save some info when in a hurry, sensing an eminent crash or an emergency, but that doesn't really make sense, does it? - If it crashes, it crashes. There's probably no time to dump anything to disk. Otherwise the crash would have been prevented.

    Anyway, some paths have been opened for when I have time to dig in a little bit deeper.

    I'll quickly find something else to annoy you ASAP. ;-)