smartd log alert

[François Patte]

Bonjour,

I have this alert in my watchlog report:

  /dev/sda [SAT] :
     Prefailure: Raw_Read_Error_Rate (1) changed to
       79, 80,
     Usage: Airflow_Temperature_Cel (190) changed to
       71, 70, 71, 70, 71, 70, 69, 70, 71, 70, 71,
     Usage: Hardware_ECC_Recovered (195) changed to
       79, 80,
     Usage: Temperature_Celsius (194) changed to
       29, 30, 29, 30, 29, 30, 31, 30, 29, 30, 29,

And same for /dev/sdb

These disks are new (I installed them this week).

What does the message mean? Do I have to send theses disks back?

Thank you.

-- 
François Patte
UFR de mathématiques et informatique
Laboratoire CNRS MAP5, UMR 8145
Université Paris Descartes
45, rue des Saints Pères
F-75270 Paris Cedex 06
Tél. [phone]
http://www.math-info.univ-paris5.fr/~patte
FSF
https://www.fsf.org/blogs/community/presenting-shoetool-happy-holidays-fr...

[Michael Schwendt]

On Sun, 27 Dec 2020 14:36:54 +0100, François Patte wrote:

>  I have this alert in my watchlog report: 
Have you used the smartctl tool from the smartmontools package yet as
to examine SMART data and perhaps run self-tests?

[François Patte]

Le 27/12/2020 à 16:14, Michael Schwendt a écrit :
>  On Sun, 27 Dec 2020 14:36:54 +0100, François Patte wrote:
>  
> > I have this alert in my watchlog report:
>  
>  Have you used the smartctl tool from the smartmontools package yet as
>  to examine SMART data and perhaps run self-tests? 
These line are from the logwatch report under the smartd section.

smartctl -x gives the same result, plus a lot of things...

-- 
François Patte
UFR de mathématiques et informatique
Laboratoire CNRS MAP5, UMR 8145
Université Paris Descartes
45, rue des Saints Pères
F-75270 Paris Cedex 06
Tél. [phone]
http://www.math-info.univ-paris5.fr/~patte
FSF
https://www.fsf.org/blogs/community/presenting-shoetool-happy-holidays-fr...

[Matthew Miller]

On Sun, Dec 27, 2020 at 02:36:54PM +0100, François Patte wrote:
>   /dev/sda [SAT] :
>      Prefailure: Raw_Read_Error_Rate (1) changed to
>        79, 80, 
As scary as this seems, this is generally normal. Raw read errors at some
fairly low rate (depending on the drive) are expected, and the new drive is
just changing from not being used at all to being in use. If you see this go
up significantly in the future, then you have a problem.

-- 
Matthew Miller
<mattdm(a)fedoraproject.org&gt;
Fedora Project Leader

[François Patte]

Le 27/12/2020 à 16:41, Matthew Miller a écrit :
>  On Sun, Dec 27, 2020 at 02:36:54PM +0100, François Patte wrote:
> >   /dev/sda [SAT] :
> >      Prefailure: Raw_Read_Error_Rate (1) changed to
> >        79, 80,
>  
>  As scary as this seems, this is generally normal. Raw read errors at some
>  fairly low rate (depending on the drive) are expected, and the new drive is
>  just changing from not being used at all to being in use. If you see this go
>  up significantly in the future, then you have a problem. 
Thank you.

What is a "fairly low rate"? And beyond which rate do I have to be scared?

Do you know if we can find somewhere what are the rates for different 
drive at the beginning of their life?


-- 
François Patte
UFR de mathématiques et informatique
Laboratoire CNRS MAP5, UMR 8145
Université Paris Descartes
45, rue des Saints Pères
F-75270 Paris Cedex 06
Tél. [phone]
http://www.math-info.univ-paris5.fr/~patte
FSF
https://www.fsf.org/blogs/community/presenting-shoetool-happy-holidays-fr...

[Matthew Miller]

On Sun, Dec 27, 2020 at 05:56:56PM +0100, François Patte wrote:
>  What is a "fairly low rate"? And beyond which rate do I
> have to be scared? 
As I understand it, it's totally manufacturer dependent and not usually well
documented. I think in general the advice is to ignore the "raw" values
unless you are diagnosing something in specific.


>  Do you know if we can find somewhere what are the rates for
>  different drive at the beginning of their life? 
I do not. :)

-- 
Matthew Miller
<mattdm(a)fedoraproject.org&gt;
Fedora Project Leader

[Michael Schwendt]

On Sun, 27 Dec 2020 17:56:56 +0100, François Patte wrote:

>  What is a "fairly low rate"? And beyond which rate do I
> have to be scared? 
Usually, you would want the displayed value to be high, since the
underlying "rate" is mapped. If the shown value drops much, that would
be reason to be concerned. The smartctl tool displays the last "worst"
value, and if it drops below the displayed threshold, such as 50% or 30%
but hardware/firmware dependent, things could go worse soon.