For the second time in 4 months im getting, seemingly out of nowhere, btrfs errors. It starts with error popups complaining that different kde applications cant write files.
After restart it doesnt boot, i tried every snapshot from the last week, all go immediately into emergency mode. Btrfs check from a live system shows tons of errors (“inode” and “transid verify failed”). No smart errors.
Could that be a hardware problem (different nvmes, so it would have to be the mainboard) or a problem with btrfs/arch?
This will be a hardware fault at some level, be it drives, mobo or even RAM. BTRFS locks the filesystem when it detects corruption to prevent future damage so it doesn’t do it “out of nowhere”. You’re going to have to look at different diag utilities or, if your area has unstable power, an UPS as a dirty shutdown can also cause this.
You say different NVMEs, but were they acquired around the same time? A while back I had 4 WD drives that started dying one after another but they were from the same batch. If this is your case, that would be the most likely explanation. Given the flash comes from like 2 companies today, even between different brands I’d be suspect.
If BTRFS or Arch were the issue, you’d be seeing a LOT more noise about it. BTRFS had some growing pains but is pretty stable at this point and most of the info about it just going nuts is fairly old at this point and not relevant.
I tested the RAM, no problems, the drives are from different manufacturers, and bought probably 1.5 years apart.
We had a power outage, but that was over a week ago. Could the corruption be unnoticed for so long?
It was going from working fine to “dead” while i was reading an article in firefox. After i tried to reboot all i could do was access the home subvolume from a live system, everything else just threw errors.
I needed to finish some work and im a bit low on storage, so i had to reinstall (on ext4) and couldnt investigate further.
Just saw this. Corruption from a power hit would show within the next reboot, it wouldn’t just suddenly crop up.
I’m still very suspect of the drives myself as a filesystem doesn’t generally nope out without a good reason (though the purchase time and all does throw a wrench in that thought). Short of verifying everything is seated (I’ve had RAM in just enough to work but loose and backed out…) this would be my next step: How to Check HDD and SSD Health with Smartmontools (2026)
Only after those give a reasonable sign of good health would I personally trust the drive again. Flukes can certainly happen, but with drives I err on the side of caution.