Comment on The right FUCKING time to get TWO ram sticks damaged

<- View Parent
Wispy2891@lemmy.world ⁨4⁩ ⁨days⁩ ago

I wonder also. I’m guessing maybe a bad lot?

The story starts two years ago when I bought them in a kit with 4 16gb sticks from micron. When installed all four the motherboard, I installed Linux and it crashed (froze) when running a VM with KVM. Tested with memtestx86 and it will always fail at the 5th test (after around 20 mins of crunching) and at reboot the bios would reset to default. Because it was an AMD Ryzen and all the web results said so, I assumed it was some kind of incompatibility and removed two sticks. With two sticks, it passed the test. I swapped the two sticks and it passed the test again. So I left those 2 16gb sticks in the Ryzen and used the other 2 16gb sticks with the Intel. Both passed the test.

Fast forward 18 months, in the Intel I’m copying a file from the nvme to the HDD and it tells me Input/output error.

I start diagnosing the btfrs filesystem, find corruption in the counter, scrub finds uncorrectable errors in the virtio-win.iso file, the one I wanted to move. I assumed it was some btrfs bug, deleted the file as I could download it again, moved on. After a few weeks a flatpak app wouldn’t start. Read the dmesg, see a btrfs message about some corrupted inode or something like that. I use find to find the file at that inode, it was the flatpak. Again assumed it was a btrfs bug, reinstalled the flatpak and moved on.

Then yesterday the system froze. This time I tested with memtestx86. It failed immediately within seconds. Took out one stick, swapped them, no change. I went back and swap them with the other two sticks bought in the same lot, those would pass the test.

source
Sort:hotnewtop