In the end the developer asked if we could turn off the screensaver. I had a quick look and couldn't be certain how I'd enabled it in the first place, so I removed 3 packages - xscreensaver mate-screensaver and electricsheep. The problem seems to have disappeared.
]]>microbug wrote:Quick update: I have replaced my motherboard and I'm still having problems. I'm now only experiencing hard power offs that leave no error messages in `journalctl`. I think this must mean that the power supply is suspect. I am ordering another power supply and will test it. Frustrating but now I know not to buy used power supplies .
Have you checked the "Power Supply Idle Control" setting in the BIOS, as a couple of people suggested?
I think I've tried that with the old motherboard but I'll test it with the new one too before I order another PSU.
]]>Quick update: I have replaced my motherboard and I'm still having problems. I'm now only experiencing hard power offs that leave no error messages in `journalctl`. I think this must mean that the power supply is suspect. I am ordering another power supply and will test it. Frustrating but now I know not to buy used power supplies .
Have you checked the "Power Supply Idle Control" setting in the BIOS, as a couple of people suggested?
]]>There are two current workarounds; switch to linux-lts kernel or add module_blacklist=ccp to your kernel command line. It sounds like the problem should be fixed in a future AGESA and/or kernel version.
]]>My next plan is to reinstall Arch on BTRFS or ZFS so that I can scrub the filesystem for corruption, as that would let me eliminate one possible cause. (I also want to upgrade from i3 to Sway so instability is a nice excuse to do that without leaving any crud from the old i3 install).
I will also try updating the BIOS if/when a new one is released.
]]>The (improved) Pastebin of journalctl is here: https://pastebin.com/d0sc0qRL.
I've downgraded to AGESA 1.0.0.2 so I'll see if that helps.
]]>FWIW I am on AGESA 1.0.0.4 so if I see further problems I'll downgrade. I'm aware of at least one issue with the latest AGESA (a timeout when starting up that causes a delay of a few seconds) that should be fixed in kernel 4.19 though. And thanks for the tip on journalctl, I didn't realise you could jump to the previous boot like that.
]]>journalctl -b-1
make sure it isn't truncated by your pager.
]]>My system was equally unstable before the upgrade. Only the motherboard and CPU were swapped, everything else is the same.
I'm not using BTRFS so I can't do a scrub. I don't have much to lose at this point — this PC is mostly used for learning more about Linux and gaming so all I need to do is back up my save files. I'll try reinstalling all packages as you suggested, then the linux-lts kernel and finally reinstalling Arch.
I'm pretty confident it's not the RAM now. Memtest86 is now on 17 completed passes with no errors, I'll leave it for 20 but I think that rules out the RAM being defective.
]]>What was your system like before the upgrade? Did you keep everything else besides the motherboard and CPU?
Maybe try the "linux-lts" kernel and see what happens.
I'm the kind of person that overclocks everything. This then means the computer crashes a lot while I experiment. I'm using btrfs, and running its "scrub" tool after crashing does rather often find corruption in files. If your PC wasn't 100% stable in the past, maybe some files are corrupted and then causing problems that would normally not happen? Maybe the hardware is fine right now and you just have to reinstall all packages to get fresh files? I don't know if there's a command or script for Arch that checks files for corruption. A script or one-liner that compares files with the contents of the packages in pacman's cache might be possible to do, so someone might have written that a script for this.
About that corruption issue, it's possible to reinstall all packages as follows, but I wouldn't run this on a computer that's not stable:
pacman -Qnq | sudo pacman -S -
(this will only reinstall Arch packages, not AUR packages)
]]>