I have a brand new Gigabyte BRIX (GB-BNI7G4-1050TI-SI-BW) with a 1TB Intel M.2 SSD which loses the ability to boot after some random time interval (hours to days). I've reinstalled this machine 7 times now. The first few installs, all would be well across several reboots, until suddenly it would only boot into purple screen. Some googling later, I followed the suggestion to install the Nvidia GPU drivers even if I don't need them (I don't, this is a headless server). That appeared to fix the problem for a few days.
So it lived for 3 days, machine was working fine, doing its work, until it suddenly crashed in the middle of the night, and went into the funny boot state all the other installs have ended up in as well: purple screen, NIC loads to where it can be pinged (DHCP), but no services. And attaching a monitor results in nothing being displayed. Can't even ctrl-alt-f1 for a login.
/var/log/syslog ends abruptly and with corrupt lines at the time when the system crashed. Each reboot so far whether it was an attempted boot into normal mode, or into recovery mode, has not written anything into the syslog. Is that normal?
During my various recovery attempts (which mostly consisted of purging and reinstalling nvidida drivers, hoping this would fix it, it didn't) grub managed to change its appearance multiple times, from once displaying grub v2.2 to now displaying v3.
I can boot into recovery mode, where I can add to the confusion by making changes to grub which then are NOT reflected after running update-grub.
This machine has an AMI EFI and when installing Ubuntu, I told the installer to write the LVM info into the EFI (force it) as there are no other OSs on this thing.
I'm quite confused as to where to even start to debug this. Ideas anyone?