Mystery crashes

This has happened twice now, a couple of weeks apart.

My desktop stops responding. My mouse cursor moves, but nothing reacts to it. The clock and system monitors on my panel freeze, and the “close window” button doesn’t highlight when I hover my mouse over it. Nothing responds. After around 30 seconds, my mother board beeps. It is a single beep, but longer than the POST check beep. It does this at random intervals. The first time I had this crash my speakers were on and I was hearing the chime of an error notification that I obviously couldn’t see just after each beep. Those panel system monitors are not telling me that my memory or CPU are overloaded - they are both at ~40% at the point of the freeze. I end up rebooting using system requests.

Both times this has happened while scrolling social media, but that could just be a coincidence - it was facebook the first time and reddit the second. My browser is Brave and my desktop is KDE. My CPU is an intel i5-7600K, and my GPU is an Nvidia geforce GTX 1050. I’m using the G05 version of the driver because the G06 version was stopping Xorg from loading.

I’m not sure if the system journal has anything relevant to say, because the crash happened at 15:12, but the last entry before my reboot at 15:15 is timestamped 15:08.

2022-09-14T15:01:10.989343+01:00 bedevere systemd[1]: snapperd.service: Deactivated successfully.
2022-09-14T15:08:17.414465+01:00 bedevere smartd[869]: Device: /dev/sda [SAT], SMART Prefailure Attribute: 1 Raw_Read_Error_Rate changed from 69 to 70
2022-09-14T15:08:17.414738+01:00 bedevere smartd[869]: Device: /dev/sda [SAT], SMART Usage Attribute: 190 Airflow_Temperature_Cel changed from 70 to 69
2022-09-14T15:08:17.414879+01:00 bedevere smartd[869]: Device: /dev/sda [SAT], SMART Usage Attribute: 194 Temperature_Celsius changed from 30 to 31
2022-09-14T15:15:13.965548+01:00 bedevere systemd[1]: Queued start job for default target Graphical Interface.

Has anyone seen anything like this before?

BIOS detecting and signalling over-temperature would be my first guess.

How old is the system? Open up the case and clean out any dust-bunnies, pay particular attention to heatsinks and ensure fans are clear of debris.

It’s about 5 years old. I used to clean it monthly, but I’ve fallen out of the habit since April

OK - Doubt it’s that then.

There’s no real standardisation of BIOS error codes AFAIK. A single long beep may (commonly) indicate a memory problem, that would be consistent with some of the symptoms you’re seeing.

Probably worth re-seating the memory.

Did that include inside the power supply? Does its fan work? Is the PSU 5 years old too? Did it come with a 5 year warranty, or only 2 or 3? It could be becoming unstable or overheating, leading to a disk read failure and resulting beep from a ns power drop that caused the reported read failure. I’d check inside the PSU for cap leaks and bulges as explained on badcaps.net and WikiPedia.

Isn’t this a little bit overshot? Why not simply consult “Motherboard Beep Codes”? Tannington had already the right hint…

That sounds just like what I deal with. I’ve narrowed the cause down to one of a few things. That exact situation happens when I open a website which tries to run some code (like special graphics) in Firefox - it’s not consistent, but it’s happened enough that I’ve been checking different settings and add-ons for Firefox. Another thing I’ve discovered is that when I turn on the system in the morning, sometimes it locks up when I start Firefox and I found that PackageKit was running in the background - but when I opened up System Monitor before Firefox and watched the resources, PackageKit didn’t show up (but it was there).

There seems to be something in the background that causes it. In my case, if I don’t open Firefox for several minutes, I rarely have a crash (usually only when I try to open up a webpage that tries to do something like start a video). I’ve checked voltages and all that - not with my scope, but using a rather good meter, I can tell that everything is running as it should.

A thermal problem with the power supply MIGHT cause it, but I found that not to be the case here.

What always fixes the problem (for a while, maybe days or up to a couple of weeks) is to use YAST first, and force updates to Firefox and your internet ‘driver’ (in my case, I use Wicked). Once that has been done, everything works great until the next time PackageKit loads on bootup and I forget to wait until it clears, or a website tries to do something fancy. It does seem that if PackageKit causes a problem, until I force an update by YAST, FIrefox gets buggier and buggier.

Hope this helps…

Bob

Over heat or out of memory maybe. I have seen FF does not like to be put in swap and uses a ton of memory. When I had 2gig it would crash if I opens FF and had a VM running. Uped it to 8 gig an all is well.