[SGVLUG] hd errors in /var/log/message

[Claude Felizardo]

Looks like your numbers are okay except for the pending remap count:

On Thu, May 15, 2008 at 6:40 AM, James Neff <jneff at tethyshealth.com> wrote:
> Thank you Claude (and others) for your response.
>
> I ran smartctl -a /dev/hda
>
> Here are the interesting lines:
>
> SMART overall-health self-assessment test result: PASSED
>
> SMART Attributes Data Structure revision number: 16
> Vendor Specific SMART Attributes with Thresholds:
> ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED
>   5 Reallocated_Sector_Ct   0x0033   253   253   063    Pre-fail Always       -       1

Does this mean it's already remapped at least once before?


> 197 Current_Pending_Sector  0x0008   253   253   000    Old_age  Offline      -       1

I believe this one means it will remap if you write to the sector.

> 201 Soft_Read_Error_Rate    0x000a   253   252   000    Old_age  Always       -       3

i guess you should monitor this and see if it starts to climb

> 203 Run_Out_Cancel          0x000b   253   252   180    Pre-fail Always       -       1

If the value (currently 253) drops to or below the threshold (180)
then that's suppose to indicate the drive may fail soon, like 24 hours
according to one web site.


Have you run the short self test to see if it reports the sector number?


> We have Spinrite so I will run that to see what it finds and if it can fix
> it.

I've used it in the past.


> I found this article too:
>
> http://www.linuxjournal.com/article/6983

Wow, that was a great article which lead me to the smartmontools FAQ
which contains a link to an article on what do do when you get a bad
block and how to force a remap!

http://smartmontools.sourceforge.net/faq.html
http://smartmontools.sourceforge.net/badblockhowto.html

It explains how to find the file that is damaged and that writing to
the sector should trigger the drive to remap. reading will keep
failing until you do this.    They use dd so it will destroy the
contents of the file.  Writing zeroes to the disk or running spinwrite
which can also rewrite every sector will trigger this.  good set of
articles.


> They mention the smartd.conf file but I don't think they answer your
> question about the DEVICESCAN option.

Looks like the man pages for smartd.conf has lots of info as well.

>
> Thanks again for the thoughtful response.
>
> --James

No prob.  It was a reminder that I've been wanting to setup smartd.

claude