Palimpsest displays out-of-range SMART values

[Gioele Barabucci]

Binary package hint: gnome-disk-utility

Palimpsest (2.28.0 on amd64) says that the "Read error rate" of my main disk is OK and has value "230874873" (raw 0xf9dec20d0000). The normalized value is "119", the worst value is "100" and the threshold value is "6".

Obviously there is something strange in these values. Maybe SMART is reporting them in the wrong way, anyway Palimpsest should do some sanity-check before reporting them.

[Gioele Barabucci (gioele)]

(I originally reported this as «https://bugzilla.gnome.org/show_bug.cgi?id=600219>)

Weird values are reported for the "Read error rate" SMART parameter for
my disk Seagate Momentus ST9250315AS with firmware revision 0001SDM1.

These values are

  value: "230874873" (raw 0xf9dec20d0000)
  normalized value "119"
  worst value is "100"
  threshold value is "6"

Could these values be sanitised somehow before reporting them?

[Lennart-poettering (lennart-poettering)]

What's wrong with that value?

Also, please attach the SMART data dump of your drive. You can generate it with 'skdump --save=foo /dev/sda'

[Gioele Barabucci (gioele)]

Created an attachment (id=30862)
Result of skdump --save on my disk

[Gioele Barabucci (gioele)]

(In reply to comment #1)
> What's wrong with that value?

I, as a simple user with no knowledge of SMART, think that the read error rate cannot be 230874873 when the threshold value is 6 and the worst value 100.

Also if that huge value is real, why are both that parameter and the disk reported as OK?

Could the fact that I am on an amd64 system influence the reading or interpretation of that value?

[Lennart-poettering (lennart-poettering)]

Nah. That field has no well-known unit or decoding, we simply show the raw unit-less value there. It's not clear if smaller or bigger is better. There's nothing wrong here.

[yonkiman (fred-yonkitime)]

I'm seeing the same thing. All 4 RAID drives and my main system RAID drive seem to be working fine, but Palimpsest is reporting 8 digit numbers for Read Error Rate and a 12 digit (both decimal) number for Seek Error Rate for all 5 disks. Not surprisingly, I occasionally get the "One of your hard disks is failing" popup as well.

When I saw the error initially I almost yanked out the disk the alert identified as "bad" and started reconstructing the array - I'm glad I looked at the other drives first!

[Sebastien Bacher (seb128)]

it's not a gnome-disk-utility bug either...

[Sebastien Bacher (seb128)]

upstream says that's not a bug

[Sebastien Bacher (seb128)]

reassigning to libatasmart since the upstream bug was moved there before being closed

[Sebastien Bacher (seb128)]

sorry about the launchpad noise should be sorted now...

[Gioele Barabucci (gioele)]

Regardless of the "ownership" of this bug, do you agree that this is a valid bug?

I am not proposing that software (libatasmart or disk-utility) should correct something that the hardware states, just that it should take care of presenting a coherent information to the user in order not to confuse her.