Rogue RAM

It was the beginning of season two of Diablo 4 when the problem first started, but I didn’t know it at the time. After updating for the new season I got warnings that my Diablo data files had been corrupted. Downloading the files fixed it and I didn’t think much of it.

Then, about a month later, I had another instance of data corruption. This time it was very large .rar files. But only the really large files were affected. Smaller things like e-books were fine. Hmmm. But a re-download would often work. I was skeptical but I moved on.

Finally, at the beginning of season 3 of Diablo 4, it happened again. Update files corrupted. I checked online and Diablo players aren’t howling so I now know it’s a me problem. The game is afoot.

When it comes to data being corrupted I usually suspect the ssd/hard drive first. So I checked the SMART status of the drives with smartctl /dev/m2x and smartctl /dev/sdx. No problems. Since both of those drives use the EXT4 filesystem I unmounted them and checked with sudo fsck -f /dev/sdxx. Fine.

Next, I ran a torture test / blend test with mprime. This tests both RAM and CPU extensively. No problems.

Ok, I really doubt the next test is necessary because it’s Linux, but let’s check for rootkits and malware. rkhunter, chkrootkit and clamscan all come out negative, false positives aside.

Huh. So I’ve got what seems to be a perfectly fine computer, that does 99% of everything I’ve asked of it without fail. But that 1% was really bugging me. So I decided to step outside the OS and boot EndeavourOS from a USB stick. Since one of the options when you boot is the low-level memtest86+ I figured, ‘why not?’.

And almost ten minutes in, there it was.

sneaky little bastard

When the first pass concluded at ~30 minutes, it was a big ole fail. FAIL

Pulling the problematic stick of RAM it passed. PASS

I got some new RAM the next day and all is well and good with my computer, but that had to have been one of the most obscure computer problems I’ve seen in a while.