Reallocated Sector Count = 67

[Robin Humble]

On Tue, Jan 05, 2010 at 09:25:24PM -0500, Stephen wrote:
> The drive is 8 months old and has 106.1 days of use. It is still under  
> warranty. It is a Seagate Barracuda ST31000528AS, firmware version CC34,  
> serial number 9VP0GXSS, current temperature 30C, last self test  
> completed OK, Self Assessment passed, 67 bad sectors.

you can probably have 1000 remapped sectors before it officially fails
itself, but in my experience any drive up in the 100's is on the way
out - it's more likely to get more dead sectors, go slow on parts of the
platters etc.
having said that remapped sectors is a pretty normal part of life these
days. on a sample of Seagate ES2 1TB disks I have handy, about 20% have
some remapped sectors after 6 months.

in smartctl -a output, look for 

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
...
  5 Reallocated_Sector_Ct   0x0033   091   091   036    Pre-fail  Always       -       196
...

where
 - VALUE starts at 100 for a new drive and decreases as you get more dead
   sectors.
 - WORST is the worst it's ever been (meaningless for this metric).
 - THRESH is the level at which the drive will self fail.

in this case, for this make/model of drive (a 1TB seagate), THRESH 036
really means ~1000-2000 remapped sectors.

the effect of having Offline_Uncorrectable/Current_Pending_Sector's is
that any attempt to read from that sector will fail, and it requires a
write to the sector to make the drive internally remap the sector (and
Reallocated_Sector_Ct will increment).

if the drive is in a Linux md raid 5/6, then md is smart enough to try
to write the missing data to a unreadable sector to force the remap.
very handy.

if anyone knows if md does this for an md raid1 then please let me know.

HTH

cheers,
robin
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists