not clear how to repair (if possible)? pool shows no errors
Device: /dev/sde [SAT], 3 Offline uncorrectable sectors.
2024-07-12 03:18:50 (America/Los_Angeles)
Run a scrub on the pool if you haven’t already
aside from that… consider replacing that disk if this system holds valuable data.
PS it’s important to know that SMART errors and ZFS errors are two different things… unrecoverable sectors are ‘fine’ in the sense that the zfs pool can self heal from them (and the drive will remap that logical sector to a different physical location of the disk). but otherwise indicate a disk that is starting to fail. unless this is an SSD then it’s probably about to fail dramatically.
1 Like
can run a SMART test on it, but sadly that usually means the drive is dead and needs to be replaced
as long as your pool is sufficiently redundant and the data isn’t super critical… you can probably live with a drive that has SMART errors.
the only gotcha is how fast is that error increasing… I had a second server with an 8x8TB RaidZ2 pool that had a ‘failing’ drive for 3+ years… it caused no harm to that system. but if this error is increasing steadily then that drive is almost certainly hosed and will be a matter of time.
of course the exception is SSDs… if you see an uncorrectable sector that usually means that the drive literally can’t garbage collect anymore and it’s actively failing.
most commonly when used with a fake/hardware raid that lacks TRIM support. (big example is Intel RST)
In case the previous answers were not clear enough:
The drive is degrading, irreversibly. You can see how bad it is in SMART reports, but you cannot fix it, except by replacing the drive. And this is exactly what you should do at your earliest convenience.
The pool shows no errors at this point: ZFS is designed for that. But one the drive fails for good, you could enter dangerous territory