Arch Linux

vectro

Member

Registered: 2016-10-15

Posts: 14

Issue with system calls, machine locking up every few days

I've been using Arch for 10+ years, and this issue has me absolutely stumped.

I have an Arch install on my Dell Precision 5810. (x86_64 Linux 5.15.71-1-lts). Using Sway/Wayland, but this bug appears in headless mode too. Note that my machine has been working fine for many years prior to this issue.

Every 2 or 3 days, my system locks up. Most applications fail to work. I am unable to successfully execute many commands (for example: sudo, systemctl, pacman, etc.). I can't even force a poweroff. My filesystems are all up and I am able to read/write files on all mount points. Upon a reboot, there are NO useful errors in journalctl -b -1 or dmesg.

I have reproduced this half a dozen times. Each time I leave a terminal running logged into root, and also leave another open with strace attached to the first terminal. When the system locks up, I can see the failing commands just halt at certain sys calls (epoll_wait(), read(), etc.. )... the failed command locks up the terminal, and I can see in strace the uncompleted sys call linger forever.

I also see nothing out of the ordinary on htop or dmesg.

I tried older and newer kernels. I tried disabling all but the bare minimum systemd services I require. Tried updating my bios.

How can I debug this system further? Could I increase the verbosity of the Kernel logs? Any recommendations are welcome. Beyond carefully looking through kernel logs and attaching strace to all my running applications, I don't know what to do.

dimich

Member

From: Kharkiv, Ukraine

Registered: 2009-11-03

Posts: 630

Re: Issue with system calls, machine locking up every few days

Check your hardware under stress test, especially memory with memtest86+.