Yesterday, my #computer started to randomly freeze.

@nemo

I have 4 memory banks, so removing 1 is no big deal. I will try by removing 1 at a time.

Overheating is not an issue. When this started, I watched the temperatur sensors, and none has ever surpassed 70°.

Thanks for the link. I didn't know the "Magic SysRQ".

@alemm

Sounds like a sensible plan.

Yesterday, my #computer started to randomly freeze.

I.e. it becomes completely unresponsive. Most of the time, the mouse cursor still moves, but clicking doesn't do anything.

Keyboard inputs have no effect whatsoever.

Video playback stops immediately, audio after a couple of seconds.

Sounds like an issue with faulty memory, or what do you think?

Edit: Turned out, it was an Nvidia driver issue.

Now fixed.

@mina

Do you had a system upgrade recently? There's an issue with GTK4.18 as I remember.

Yesterday, my #computer started to randomly freeze.

I.e. it becomes completely unresponsive. Most of the time, the mouse cursor still moves, but clicking doesn't do anything.

Keyboard inputs have no effect whatsoever.

Video playback stops immediately, audio after a couple of seconds.

Sounds like an issue with faulty memory, or what do you think?

Edit: Turned out, it was an Nvidia driver issue.

Now fixed.

I have this issue as well. I replaced most of my hardware since it started but it's still happening.

@mina

Yesterday, my #computer started to randomly freeze.

I.e. it becomes completely unresponsive. Most of the time, the mouse cursor still moves, but clicking doesn't do anything.

Keyboard inputs have no effect whatsoever.

Video playback stops immediately, audio after a couple of seconds.

Sounds like an issue with faulty memory, or what do you think?

Edit: Turned out, it was an Nvidia driver issue.

Now fixed.

@mina or memory exhaustion. I have seen similar behaviour, especially when the swap partition was on an encrypted partition as then every read from swap has to go through the decryption but before that, the evicted page has to be encrypted. I have observed load values >70 in those situations. After a while, oomkiller will terminate a process, quite often the browser, and the system will become responsive again.
Since most newer distributions have moved to zram for swap, this has become rarer

@pjakobs

I don't have any encrypted partitions and with 32GB of RAM, it's hard to exhaust it, even with modern websites, I guess.

@mina I've seen it happen, albeit with uptimes >50 days and heavy Firefox use, often way more than 50 tabs.
Have a terminal open monitoring top/btop when it happens again.
Also, if you suspect hardware, have a look at dmesg

@pjakobs

How can you live with so few tabs?

If taking memory banks out doesn't solve the problem, I will start the detective work.

@mina you can also run `journalctl --boot -1` and see if there are any oops'es or other obvious failures.
Generally, I found that dropping a log like that into one of the typical llms can give a pretty good first hin to what possible issues might be contained in it

@pjakobs

Sounds like a good idea.

Those logs have so many entries, and with most, I have no idea what they mean.

@mina oh, ich verwechsel das immer: --boot -1 ist der vorletzte Boot, --boot 0 der letzte (aktuelle).

für Memory Issues ist auch
journalctl -ke | grep -i "out of memory" hilfreich, eventuell kombiniert mit --boot -1 wenn Du gerade nach einem solchen Hänger neu gestartet hast

@mina wenn das Problem häufig auftritt kann es auch hilfreich sein, eine Shell mit `sudo dmesg -wT` offen zu haben - das sind die live Kernel Messages - wenn der Kernel bemerkt, dass was nicht stimmt, dann wird das hier sichtbar.

@mina wenn das Problem häufig auftritt kann es auch hilfreich sein, eine Shell mit `sudo dmesg -wT` offen zu haben - das sind die live Kernel Messages - wenn der Kernel bemerkt, dass was nicht stimmt, dann wird das hier sichtbar.

@pjakobs

journalctl -ke --boot -1

hat's gebracht.

Das Problem war (was sonst) die Nvidia GPU. Die ist gecrasht.

Ich installiere gleich mal den aktuellen Treiber.

@mina ah, ja, die sind immer problematisch,vor allem, wenn es solche mit binärem Treiberteil sind - ich glaube auch für nvidia gibt's heute open source Treiber, oder?

aber wie auch immer - schön, dass Du es finden konntest.