From: Franc Zabkar on
On 22 Jan 2010 09:43:07 GMT, Arno <me(a)privacy.net> put finger to
keyboard and composed:

>Well, I think I still do not undertsand the seek error rate attribute,

Here are two examples for two Seagate ST3500630AS drives:

ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE

7 Seek_Error_Rate 0x000f 035 024 030 Pre-fail
Always In_the_past 54443054162961

7 Seek_Error_Rate 0x000f 059 048 030 Pre-fail
Always - 1507770244954


Convert the raw decimal attribute values to hex:

54443054162961 = 0x318402e76411 (drive A)
1507770244954 = 0x015f0e1c1f5a (drive B)

AIUI, the number of seek errors is stored in the uppermost 16 bits of
the 48-bit attribute value, and the total number of seeks is stored in
the lower 32 bits.

So for drive A,

# seek errors = 0x3184
# total seeks = 0x02e76411

The normalised attribute value is logarithmic and appears to be
calculated as follows:

Normalised value = -10 log (seek errors/total seeks)

If the number of seek errors is 0, then use a value of 1.

For drive A, this works out as ...

-10 x log(0x3184 / 0x02e76411) = 35.8471493

For drive B it is ...

-10 x log(0x015f / 0x0e1c1f5a) = 58.2893528

Not exactly right, but very close ...

The worst case value of 24 corresponds to an error rate of ...

-10 ^ 2.4 = 1/251

.... ie one error in every 250 seeks.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.
From: Franc Zabkar on
On Tue, 26 Jan 2010 10:39:03 +1100, Franc Zabkar
<fzabkar(a)iinternode.on.net> put finger to keyboard and composed:

>The worst case value of 24 corresponds to an error rate of ...
>
> -10 ^ 2.4 = 1/251

Sorry, that should be ...

10 ^ (-2.4)

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.
From: Arno on
In comp.sys.ibm.pc.hardware.storage Franc Zabkar <fzabkar(a)iinternode.on.net> wrote:
> On Tue, 26 Jan 2010 10:39:03 +1100, Franc Zabkar
> <fzabkar(a)iinternode.on.net> put finger to keyboard and composed:

>>The worst case value of 24 corresponds to an error rate of ...
>>
>> -10 ^ 2.4 = 1/251

> Sorry, that should be ...

> 10 ^ (-2.4)

> - Franc Zabkar
> --
> Please remove one 'i' from my address when replying by email.

Thanks, noted and archived in my doc collocetion.

Arno

--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: arno(a)wagner.name
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans