I have been using a company computer running Ubuntu 22.04. There are frequent and unexplained problems, like segmentation faults, stack errors, files disappearing, computer freezing or not booting, or turning off immediately after I turn it on. I don’t know what to do. The IT staff came to my office to check the computer and said “it was all good.” I am not allowed to boot from a USB stick or enter BIOS or open the case. I ran a command line memory check several times with no errors. There is an NVIDIA card, but it’s running X.org and usually headless. I mostly set up tasks via SSH.

What would you do?

  • TimeSquirrel@kbin.melroy.org
    19·
    6 months ago

    Segmentation and stack errors are most certainly bad memory, I’m 99% sure of it, reboot and run mem test from GRUB if you have the option. The “stack” is the non-dynamically allocated space your program is assigned to run in. Stack errors mean some pointers somewhere are likely getting corrupted and it’s trying to access addresses beyond what it’s allowed to access.

    • LogarithmicCamel@feddit.ukOP
      21·
      6 months ago

      I can’t run memtest unfortunately. The option isn’t there and I don’t have permission to boot from a USB stick.

      • Avid Amoeba@lemmy.ca
        6·
        6 months ago

        If you have root you could theoretically add Memtest86+ to the boot order. There’s tools that allow adding boot entries in EFI. You could probably place a Memtest86+ binary in your EFI partition and register it with the EFI firmware. But I’m not suggesting to do it since you could make the machine unbootable and the problem might be on the storage path. I’m just thinking of should be possible.

        • LogarithmicCamel@feddit.ukOP
          7·
          6 months ago

          I can sudo. Last time I looked into this, Memtest86+ version 6 was required to work with UEFI but it wasn’t available for Ubuntu 22.04. Now it seems that 24.04 has it, so I might update and see if I can get the test running. Thanks for the suggestion!

          • Avid Amoeba@lemmy.ca
            2·
            6 months ago

            You can get the binary from the project’s website. Still not suggesting to f around with it.

            • Dave.@aussie.zone
              3·
              6 months ago

              They should be able to put memtest on the boot partition and then break to an EFI shell on boot and Ioad it manually.

              There will be a bit of swearing and googling required but it’s doable in a way that doesn’t mess with the current boot arrangement.

      • exu@feditown.comEnglish
        1·
        6 months ago

        You could download stressapptest and run that memory benchmark in the normal system.

        I’m not sure how well the current version of Memtest does, but when I was overclocking I was told not to use it as it couldn’t reliably get memory to crash. (Funny problem to have). The two recommended tools are Windows only, so I found stressapptest as the best alternative.

  • Strit@lemmy.linuxuserspace.show
    15·
    6 months ago

    As it’s running Ubuntu you could provide your IT department with the logs from the crashes, so they can see there is a problem.

    If they provided the Ubuntu install it’s their job to support it.

  • NegativeLookBehind@lemmy.worldEnglish
    7·
    6 months ago

    Do you have a way to reproduce the problem so IT can see it? Have you taken screenshots (or just pictures with your phone) of the problem if it’s hard to reproduce?

    • LogarithmicCamel@feddit.ukOP
      5·
      6 months ago

      It’s not hard to reproduce, but it’s annoying that when they finally came here to check it, no problems happened. I had to bug them so much to even get them to have a look.

  • absGeekNZ@lemmy.nzEnglish
    3·
    6 months ago

    I was having a lot of random crashes and weird errors on my Mint install, using the logs, I tracked it down to a SSD fault.

    I really didn’t want to send it back, since I got it from Amazon and I’m in NZ… So after a bit of checking I found that the FW on the SSD was not the latest. Updated the FW, went from at least 1 crash per workday, to no crashes in the last 6 months.

    My SSD is a WD SN850X 4TB