SSD reports many reallocated sectors

[Japplo]

Hi Guys,
yesterday smartctl reported, that 14336 sectors are reallocated:

smartctl -a /dev/sda | grep Reallocated_Sector
5 Reallocated_Sector_Ct 0x0033 100 100 000 Pre-fail Always - 14336

I'm a little bit surprised... Is that true or is it a Bug in F15?

[sej7278]

i thought you weren't supposed to use smart with ssd's as it could damage them.

[Japplo]

Why should smart damage the ssd? I've never heard that before... I've run a BIOS HDD check without problems. Is it possible to reset the smart counters?

[tuxor]

I'm having 0 sectors with my toshiba ssd.
# smartctl -a /dev/sda | grep Reallocated_Sector
5 Reallocated_Sector_Ct 0x0013 100 100 050 Pre-fail Always - 0

[Japplo]

All errors occurred at the same time. I cannot believe that this is true.

[stevea]

PLEASE POST WHICH DRIVE YOU ARE USING - I AM CURIOUS

SMART technology is built into SSD controllers - so obviously it won't generally damage an SSD. Duh !

The variables kept for SMART technology were designated for rotating drives and don't perfectly map into SSD technology, so there is some oddness on what you get from SMART on an SSD. Also most SSD drivers aren't in the smartctl database ,limiting the value. Try "smartctl -i ..." and look for:
Device is: Not in smartctl database [for details use: -P showall]


Each type of SSD controller and even firmware revs can give different results. So we shouldn't expect that an Intel, a Crucial and a SandForce controllers will report SMART info similarly. It might make sense to report your drive make and firmware version.

There is no method of resetting SMART counters. It's a bit like rolling back your odometer, frowned upon. I'm sure the OEMs can do it. You may want to run the extended tests on the disk

You should look into upgrading the firmware on you SSD, if newer firmware is available.

---------- Post added at 08:52 AM ---------- Previous post was at 08:33 AM ----------

===============
Interesting factoid. Your bad sector count is 14336 = 7 * 2048 sectors and amount to exactly 7MB (512B sectors)
That's not some accumulated bunch of errors, it's a nice round number in binary.

My *guess* is that during garbage collection your SSD controller found 7MB of pages (an erase page is usually in the range of 0.25MB to 1MB) that would not erase cleanly and it marked these pages with SMART "Reallocate Sector",

You should understand that all SSD devices are made up of FLASH .devices with limited lifespan. The current chips are shilled with up to 2% uncorrected errors and ~3% extra cells where these errors and others that develop over the chip lifespan must be corrected with software(firmware). The drive controller manufacturers create various error correction schemes to manage this.

PLEASE POST WHICH DRIVE YOU ARE USING - I AM CURIOUS

[Japplo]

Hi stevea,
thank you for the informations, I use a Crucial C300 with the newest firmware and the drive is not in the smartctl database. In the last time, I was forced to reset my notebook during some problems with F15. I believe that the problems occurred during a reset but I'm not 100% sure.
____

I've found similar problems in the crucial forum http://forum.crucial.com/t5/Solid-State-Drives-SSD/bd-p/ssd . It seems, the smartcrl will not show the correct data

[tuxor]

NOT IN SMARTCTL DATABASE and 0 sectors though!

# smartctl -i /dev/sda
smartctl 5.40 2010-10-16 r3189 [x86_64-redhat-linux-gnu] (local build)
Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net

=== START OF INFORMATION SECTION ===
Device Model: TOSHIBA THNS128GG4BAAA-NonFDE
Serial Number: 99RS1023T02Z
Firmware Version: AGLA0203
User Capacity: 128.035.676.160 bytes
Device is: Not in smartctl database [for details use: -P showall]
ATA Version is: 8
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Sat Mar 26 14:20:35 2011 CET
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

# smartctl -a /dev/sda | grep Reallocated_Sector
5 Reallocated_Sector_Ct 0x0013 100 100 050 Pre-fail Always - 0

[Japplo]

-> http://forum.crucial.com/t5/Solid-State-Drives-SSD/Crucial-REALssd-2048-bad-sectors-according-to-SMART-data/m-p/39882/highlight/false#M12640

PLEASE POST WHICH DRIVE YOU ARE USING - I AM CURIOUS
SMART technology is built into SSD controllers - so obviously it won't generally damage an SSD. Duh !

OK.... But why is the /usr/sbin/smartd -n -q never process active, if I restart the computer? I can't find it with chkconfig --list.

[stevea]

Japplo,

Crucial C300 is a top notch product. made by MicronTech which makes the FLASH parts. They do use a proprietary controller (one of the fastest) so there isn't much info out there yet. BTW If I was in the market I'd but Crucial SSD too.

My *guess* is that you are seeing an actual drive error that has been corrected and are not seeing an artifact of the F15 update. The OS simple doesn't have the ability to mess up the SMART data variables. As SSDs age they will develop errors - they only have a limited lifespan. The "special sauce" in an SSD drive is how robust is the error correction. I think you are OK. It's unfortunate that a (correctable) error occurred in the first year of life - but more errors will surely follow for all of us SSD users.

So long as the 'smartctl -H' passes then the drive is fine.

Good idea to post to the crucial forum.

====
Tuxor,

# smartctl -a /dev/sda | grep Reallocated_Sector
5 Reallocated_Sector_Ct 0x0013 100 100 050 Pre-fail Always - 0

The '0' at the end of the second line means that NO sectors are reallocated. You have no problem.
The other numbers are ranges of acceptable values (very hard to interpret - you need to read the
SMART docs) before the drive is considered failed - but you don't have any problem with realloc.

[sej7278]

from /lib/udev/rules.d/95-devkit-disks.rules :

# ATA disks driven by libata
# These are disabled by default, since they can cause hardware damage on some
# SSD disks. Please see https://launchpad.net/bugs/445852 for details.
KERNEL=="sd*[!0-9]", ATTR{removable}=="0", ENV{ID_BUS}=="ata", ENV{DEVTYPE}=="di
sk", IMPORT{program}="devkit-disks-probe-ata-smart $tempnode"

on ubuntu at least they disable smart in palimpset