Random full system freeze

Whoracle · 2025-08-04 09:45:19

Hello everyone

for quite a while now I'm experiencing full system freezes at random. The only common cause I can find is that it's (almost always) when I'm in an active Teams Meeting in Vivaldi. The whole system just locks up - No more audio, no reaction to keypresses at all, no reaction to CTRL+ALT+DEL, only a poweroff via the power button does anything. Can't switch ttys, I just have the last frame on-screen and that's it.

This doesn't happen predictably - sometims I get 3 of these in a day, sometimes two in a row within minutes of each other, sometimes it's days between crashes.

I don't think it's load, because I do not experience any lock ups when stress testing the system or playing a game or whatever.

journalctl -b -1 right after the crash doesn't show anything:

Aug 04 11:13:20 lynxcore sudo[50920]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan
Aug 04 11:13:20 lynxcore sudo[50920]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50920]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:20 lynxcore sudo[50924]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme
Aug 04 11:13:20 lynxcore sudo[50924]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50924]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:20 lynxcore sudo[50927]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:20 lynxcore sudo[50929]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:20 lynxcore sudo[50928]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:20 lynxcore sudo[50930]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:20 lynxcore sudo[50927]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50929]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50928]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50930]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:20 lynxcore sudo[50928]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:20 lynxcore sudo[50927]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:20 lynxcore sudo[50930]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:20 lynxcore sudo[50929]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50940]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan
Aug 04 11:13:30 lynxcore sudo[50940]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50940]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50945]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme
Aug 04 11:13:30 lynxcore sudo[50945]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50945]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50951]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:30 lynxcore sudo[50950]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:30 lynxcore sudo[50949]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:30 lynxcore sudo[50948]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --info --health --attributes --tolerance=verype>
Aug 04 11:13:30 lynxcore sudo[50951]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50950]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50949]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50948]: pam_unix(sudo:session): session opened for user root(uid=0) by (uid=959)
Aug 04 11:13:30 lynxcore sudo[50948]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50951]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50949]: pam_unix(sudo:session): session closed for user root
Aug 04 11:13:30 lynxcore sudo[50950]: pam_unix(sudo:session): session closed for user root

Full journal from journalctl -b -1 after the last crash: https://0x0.st/8hsn.txt

I'm logging temps for both my GPU and the rest of my system via telegraf to my homelab, and I don't see a rise in temps over the telegraf interval (10 seconds), so if it's a thermal thing, it happens FAST. See this image of the minute during which the crash occured - metrics collected by telegraf and sent to remote server: https://img.lynxcore.org/250804_114222.png

Specs:
AMD Ryzen 7800X3D
NVidia RTX 3080FE
64GB Corsair Vengeance DDR5@6200MT/s

Most likely useless NeoFetch output:

OS: Arch Linux x86_64 
Kernel: 6.15.9-arch1-1 
Uptime: 24 mins 
Packages: 1929 (pacman), 19 (flatpak) 
Shell: zsh 5.9 
Resolution: 1200x1920, 3440x1440, 2560x1440 
WM: awesome 
Theme: Adwaita-dark [GTK2/3] 
Icons: Obsidian-Mint [GTK2/3] 
Terminal: urxvt 
Terminal Font: Source Code Pro 
CPU: AMD Ryzen 7 7800X3D (16) @ 5.053GHz 
GPU: NVIDIA GeForce RTX 3080 
GPU: AMD ATI 0d:00.0 Raphael 
Memory: 12218MiB / 63427MiB

amdgpu blacklisted because I don't use the on-board GPU.

Things I didn't yet try:
- REISUB/SysRq
- SSHing into the machine

Because I'm in active meetings when this occurs I usually don't have time to try those two. Will try when it happens again in one of the more useless meetings.

Definitely open programs when this occurs:
- 1 Instance of Vivaldi running the teams meeting

Usually, but not always open programs:
- 1 additional Vivaldi instance
- 1 instance of google chrome
- 1 or 2 instances of rxvt-unicode
- sublime-text
- sometimes I'm connected to the company VPN via OpenVPN, but not always

Any pointers welcome.

Last edited by Whoracle (2025-08-04 10:13:58)

xerxes_ · 2025-08-04 15:21:41

Do you use Vivaldi from Arch repo or from AUR or from elsewhere? Did you tried run Teams Meeting in other browsers (Chromium, Firefox) from Arch repo?

Whoracle wrote:

Things I didn't yet try:
- REISUB/SysRq
- SSHing into the machine

Definitely try!

Also see: https://wiki.archlinux.org/title/Ryzen

Whoracle · 2025-08-04 15:42:32

Vivaldi from extra. I had the same issue in Chromium and Chrome a few months back before I switched my work profile over to Vivaldi. Never tried Firefox, don't have that installed. Will go over the Ryzen article, but I doubt I'll find the solution in there since the issue is so... specialized? But I'll see.

Whoracle · 2025-08-06 14:20:35

Happened again just now. SSH did not work, and neither did REISUB. Looks like my wireless KB didn't even get through to the dongle anymore.

journalctl -b -1: http://0x0.st/8hWo.txt

seth · 2025-08-07 20:53:49

…
Aug 06 16:13:10 lynxcore sudo[124790]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme
Aug 06 16:13:20 lynxcore sudo[124807]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan
Aug 06 16:13:20 lynxcore sudo[124811]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme
Aug 06 16:13:30 lynxcore sudo[124846]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan
Aug 06 16:13:30 lynxcore sudo[124850]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme
Aug 06 16:13:40 lynxcore sudo[124870]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan
Aug 06 16:13:40 lynxcore sudo[124874]: telegraf : PWD=/ ; USER=root ; COMMAND=/usr/bin/smartctl --scan --device=nvme

wtf is that?
https://aur.archlinux.org/packages/telegraf-bin

https://wiki.archlinux.org/title/S.M.A. … self-tests

Disable that, just to reduce noise and interference.

Next to https://wiki.archlinux.org/title/Ryzen#Troubleshooting you might be facing https://wiki.archlinux.org/title/Solid_ … leshooting

If you can somewhat trigger this, make sure to keep a visible terminal emulator running dmesg -w on screen.
Maybe it allows you to catch some last messages from the kernel before things go south.

Whoracle · 2025-08-08 06:15:31

seth wrote:

wtf is that?

SMART plugin for telegraf. Disabled now.

Next to https://wiki.archlinux.org/title/Ryzen#Troubleshooting you might be facing https://wiki.archlinux.org/title/Solid_ … leshooting

Possible, but I'd expect to see some of the mentioned events in some log or other, no? (Also, I have none of these vendors' drives - mine are all Crucial. Doesn't mean much, granted).

If you can somewhat trigger this, make sure to keep a visible terminal emulator running dmesg -w on screen.
Maybe it allows you to catch some last messages from the kernel before things go south.

100% not reproducible at will, but I've got enough monitors to keep a dmesg open at all times, so I'll try. I think I remember having one open a while back, when I was getting 3 crashes in a row, and it not displaying anything, but worth making sure and documenting the outcome here.

Gonna plug in an additional, wired keyboard for REISUB, too.

seth · 2025-08-08 06:57:38

I'd expect to see some of the mentioned events in some log or other, no?

You'd only get the MCE errors on spontaneous reboots, everything else will be lost w/ the power button.

SMART plugin for telegraf. Disabled now.

Just to be clear, the comment was rather on the "interesting" implementation here - I wasn't faulting you for using it (and actually knew where this is kinda coming from)
I was just too tired and baffled for a more measured comment than "wtf"

Whoracle · 2025-08-08 07:00:28

No worries. I'm not happy with it either, but it's the only way I have found to get the disk fan speed and temps into influx.

Now we wait another few days for the next crash to happen.

Arch Linux

#1 2025-08-04 09:45:19

Random full system freeze

#2 2025-08-04 15:21:41

Re: Random full system freeze

#3 2025-08-04 15:42:32

Re: Random full system freeze

#4 2025-08-06 14:20:35

Re: Random full system freeze

#5 2025-08-07 20:53:49

Re: Random full system freeze

#6 2025-08-08 06:15:31

Re: Random full system freeze

#7 2025-08-08 06:57:38

Re: Random full system freeze

#8 2025-08-08 07:00:28

Re: Random full system freeze

Board footer