My current tower started out on Windows, and for some reason after a year or so it started crashing out randomly. Load didn’t matter, it would pass benchmark tests and then crash randomly 5mins after boot. However there was not a single useful error I could find. Installed Fedora, and looked at journalctl after a crash. Immediately I see “GPU has fallen on the bus”. Apparently it is relatively common, but I also found a thread that said it actually can be caused by loose connection. Did a complete reinstall on my GPU, haven’t had the problem again (~6mo now, had both 535 and 545 drivers). Sometimes it really might be a descriptive error message 😆
Fresh from the box OEM hardware, either Dell or HP... I forget. A laptop system.
We couldn't get the damn thing to do windows updates, which was part of our initial prep for the system. It kept crashing, no useful info from logs.
I booted off of an Ubuntu live image I had on a USB. Turns out, one of the CPU cores on die was faulty. If I reduced the CPU cores visible to the system to one, it worked fine. All enabled? Crash.
A call to support and a quick service visit sorted out the system, but Microsoft's error reporting was useless at diagnosing the issue.
I'm just trying to figure out why, if you'd programmed it to detect /boot wasn't mounted properly, instead of having it error out, you'd have it do the install anyway and then snarkily print "hey, jsyk, that thing I just did probably led to system corruption. Hope you're reading this log and/or feel like spending five hours debugging a nonbootable system! Have fuuuun!"