Re: 1 Currently unreadable (pending) sectors How worried should I be?
On 2 Jan 2024 20:17 -0700, from charlescurley@charlescurley.com (Charles Curley):
> Jan 02 20:07:39 tiassa smartd[740]: Device: /dev/sda [SAT], 1 Currently unreadable (pending) sectors
This is not the problem. This is smartd reporting something about the
drive's health which you might be interested in. (Also, about what
someone else wrote, it's not really surprising if smartd checks the
drive every 30 minutes. It would have been more curious if there were
kernel I/O errors logged exactly every 30 minutes, but you haven't
shown anything from those logs in this thread AFAICT.)
What I find curious is the combination of Reallocated_Sector_Ct == 1
and Reallocated_Event_Count == 0. There's also the
Current_Pending_Sector == 1 but Offline_Uncorrectable == 0 even after
two SMART health tests, one of which being an extended offline test.
If a sector has been reallocated, that should have happened at some
point, so if Reallocated_Sector_Ct > 0 then Reallocated_Event_Count
_should_ also be greater than 0 (and hopefully not greater than
Reallocated_Sector_Ct), which it isn't reported as in your case.
Likewise, after an extended offline SMART test, each sector should
have a known status of either readable or not readable. If the
firmware detects a sector as being marginal, it _should_ rewrite it
and check again; if it's still marginal, it _should_ reallocate that
sector, which _should_ increment Reallocated_Event_Count. The "pending
sectors" SMART attribute is supposed to count sectors which the drive
has failed to read, so they cannot be reallocated, and which will be
reallocated on the next write (when the drive knows what data to put
in the reallocated-to sector). Since both tests finished without
finding any errors, there _should_ have been no unreadable sectors.
I'm inclined to believe that your drive is fibbing SMART data.
As a background process, try running something like
# ionice find / -xdev -type f -exec cat {} + >/dev/null
and if that doesn't cause any I/O errors to be output or logged, then
the drive is _likely_ fine. (You may need to adjust for other file
systems also on that drive, such as /boot.)
--
Michael Kjörling 🔗 https://michael.kjorling.se
“Remember when, on the Internet, nobody cared that you were a dog?”
Reply to: