systemDiscourse #3 – Dead RAM, Funky UEFI, and Reinstalling Arch
Another month, another systemDiscourse.
It’s been an interesting few weeks for my workstation. I’ve done some minor upgrades, had some minor panic, and wiped the slate clean.
I. System Upgrade Saga: UEFI Shenanigans
So for this round of upgrades, I got a new motherboard, new CPU, some more RAM, and moved everything over to a new case. The build itself went smoothly enough in the monstrously spacious Phanteks Enthoo Pro and it was off to the races. Then the first sign of trouble showed itself.
As far as I know, it’s usually not necessary to reinstall Linux when switching platforms. Not even with Windows, though it used to be more of a pain point in years past. I thought everything would just work but upon booting up and configuring my UEFI settings, I quickly realized this MSI Tomahawk motherboard did not see GRUB at all. This lead to the first bout of frustration on my end. I booted off Arch installation medium, and got to work.
After a decent bit of effort with regenerating my GRUB config and messing with my motherboard settings, I decided to just go ahead and nuke GRUB. I’ve never used another bootloader for Linux before but if this wasn’t working, now was a good enough time as any to give it a shot. Since this system was purely UEFI and I don’t really have any reason to legacy boot my system, I decided to first try rEFInd. After chrooting into my main drive, nuking my EFI partition, and successfully configuring rEFInd, I did another reboot. I didn’t see rEFInd anywhere in my boot options either. I went ahead and reinstalled and reformatted my EFI partition a few more times just for the hell of it to no avail.
As a last-ditch effort I tried to go a different route and just use plain old systemd boot. After wiping my EFI partition once again and setting up my /boot directory, I gave my machine one last reboot. Like magic, I could finally see a proper boot option through my motherboard.
This was only the beginning, though, as the next few days would begin to test me.
II. Bad RAM and Arch Reinstall
Like the title suggests, the RAM part of this upgrade would prove to be a pain point. I had no idea just yet, though, and through a combination of lacking sleep and ignorance it would be quite a ride till I figured this out.
With my system built, a functioning bootloader installed, and everything operational, I brought my behemoth of a computer back up to my desk and hooked everything up. At first it seemed like it was working flawlessly. On the software side, everything was where I left it. On the hardware side, things were looking functional. Then I started using the thing.
Within the first night and day of using my system, a few more signs of trouble showed themselves. Running system upgrades through pacman would sometimes just fail to output anything. I would have to kill my terminal and run the command again for it to work. Firefox tabs would randomly crash, showing a crash thing that I had never had the misfortune of being shown before on this install. I would game for a while without issue, then the game would crash, or my entire system would lock up. For some reason this seemed to be mostly happening with CS:GO but I was also playing it the most at the time, so I’m guessing the same would’ve happened with other games given enough time.
I spent a while troubleshooting the software side of things finding no answers. This was my first mistake: blaming the OS install despite having just done a hardware upgrade. In the back of my head, I was already considering reinstalling Arch, though, before encountering any problems. I’ve had this install going since mid-2019 and it was my first Arch setup ever. There were so many abandoned files, broken configs lingering in obscure directories, and probably other bad habits that left their mark as I learned how to properly configure and maintain this DIY system. I thought a clean slate would be a good option and would be simple enough as my home partition was already separated.
Clean install. Reboot. Enter new environment. More trouble…
Pacman commands continued to fail. Once everything was setup, Firefox would continue crashing. Games would continue to lock up my system. I thought back to the bootloader fiasco and thought maybe it had something to do with the motherboard since it’s already given me some trouble. I decided to flash my mobo firmware and see if an update would fix the stability at all. After a brief moment of panic post-update, having to boot without my GPU installed, then reboot with my GPU back in to get my display output working again, I was back home. After some testing and hopefulness, it started happening again.
I don’t know how long it took me to come to the correct conclusion but I eventually did. I checked around for good options to stress test a system for issues and decided using memtest86 would be my first option.
Well, a few tests went smoothly but halfway through the pass I was flooded with errors. It was bad RAM. This whole time, it was one of the 2 new DIMMs of RAM. I decided to coin toss and take out 2 of the sticks at the same time. Tested again, perfect score for multiple passes. No errors at all.
This was kind of shocking to me. I’ve actually never had a dead on arrival part in my building days yet. I’ve put together a handful of systems and gone through a handful of upgrade cycles. Not once did a new part fail me. I suppose, though, that bad RAM is more common than other types of dysfunctional parts. It’s much better than the CPU or mobo being bad, too. Oh well, the saga is over, my system is functional, and I have a fresh new Arch install to play with.
With everything i need set up and my apps reinstalled, I have a functional, usable system with new configs (i cleared my .config directory in home too, just to start extra fresh), and all with a grand total of 1337 packages installed, on the dot. That has to be a sign of great things to come.
Anyways, that’s it for this entry. Catch me next month when I nuke my Arch install once more and convert to Windows 11!
(jk i could never)