headers
mathog | 13 Jun 2012 17:29
Picon
Favicon

pending blocks not reliably found in -t long

This device

Device Model:     WDC WD10EARS-00Y5B1
Firmware Version: 80.00A80

picked up some pending sectors.  Strangely these are not always found 
with a
smartctl -t long /dev/sda

(smartmontools release 5.41 on Mandriva 2010 x86_64)

That is, before the test starts there are N pending, after the test 
ends there are N pending,
but no problems are found, even though the pending sector count would 
indicate a problem.
Here are the last 3 logs,  the first one found a pending sector (there 
were 17 at the time)
and #2 caused one more pending sector to be set - but didn't find it.  
Neither did a subsequent
long test (#1).

# 1  Extended offline    Completed without error       00%     19531    
     -
# 2  Extended offline    Completed without error       00%     19525    
     -
# 3  Extended offline    Completed: read failure       50%     19505    
     973466800

I'm currently scanning the disk with:
(Continue reading)

Permalink | Reply |
headers
Rich | 15 Jun 2012 00:42
Favicon

Re: pending blocks not reliably found in -t long

I wouldn't expect a long SMART test to necessarily find sectors it
already knows are pending reallocation on next successful read or
overwrite, nor would I call one or two sectors pending reallocation a
reason to predict failure.

It's also not a failure in smartmontools, since the internals of how a
SMART short/long test are implemented are drive-specific, smartctl
just triggers the run.

As far as the blocks not being removed, are you sure you overwrote the
whole blocks?

Linux hanging on an unmount is going to be specific to the filesystem,
version of Linux, [...]

- Rich

On Wed, Jun 13, 2012 at 11:29 AM, mathog <mathog <at> caltech.edu> wrote:
> This device
>
> Device Model:     WDC WD10EARS-00Y5B1
> Firmware Version: 80.00A80
>
> picked up some pending sectors.  Strangely these are not always found
> with a
> smartctl -t long /dev/sda
>
> (smartmontools release 5.41 on Mandriva 2010 x86_64)
>
> That is, before the test starts there are N pending, after the test
(Continue reading)

Permalink | Reply |
headers
Weedy | 18 Jun 2012 04:41
Picon

Re: pending blocks not reliably found in -t long

On 13/06/12 11:29 AM, mathog wrote:
> I'm currently scanning the disk with:
> 
>    dd if=/dev/sda7 of=/dev/null bs=4096

http://hdrecover.sourceforge.net/

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Permalink | Reply |
headers
Christian Franke | 18 Jun 2012 07:09
Picon
Favicon

Re: pending blocks not reliably found in -t long

Weedy wrote:
> On 13/06/12 11:29 AM, mathog wrote:
>> I'm currently scanning the disk with:
>>
>>     dd if=/dev/sda7 of=/dev/null bs=4096
> http://hdrecover.sourceforge.net/
>

or:

http://www.gnu.org/software/ddrescue/ddrescue.html

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Permalink | Reply |
headers
Christian Franke | 18 Jun 2012 07:38
Picon
Favicon

Re: pending blocks not reliably found in -t long

mathog wrote:
> This device
>
> Device Model:     WDC WD10EARS-00Y5B1
> Firmware Version: 80.00A80
>
> picked up some pending sectors.  Strangely these are not always found
> with a
> smartctl -t long /dev/sda

Recently I had a similar issue with a SAMSUNG HE103UJ: 40 pending 
sectors were reported but both SMART long test and a read scan with 
ddrescue didn't report any bad sectors. The pending sector count was 
reset after zeroing each disk block with ddrescue.

Apparently this firmware resets the pending status only on write even if 
the read error was transient. This is vendor specific and typically 
never documented.

Thanks,
Christian

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Permalink | Reply |
headers
mathog | 18 Jun 2012 18:44
Picon
Favicon

Re: pending blocks not reliably found in -t long

On 17-Jun-2012 22:38, Christian Franke wrote:
> mathog wrote:

>> picked up some pending sectors.  Strangely these are not always 
>> found
>> with a
>> smartctl -t long /dev/sda
>
> Recently I had a similar issue with a SAMSUNG HE103UJ: 40 pending
> sectors were reported but both SMART long test and a read scan with
> ddrescue didn't report any bad sectors. The pending sector count was
> reset after zeroing each disk block with ddrescue.
>
> Apparently this firmware resets the pending status only on write even
> if the read error was transient. This is vendor specific and 
> typically
> never documented.

How did you find the bad blocks if subsequent reads completed?

Also bizarre on the WD10EARS drive was what happened when the bad 
blocks were overwritten, this is the final state:

   5 Reallocated_Sector_Ct   0x0033   200   200   140    Pre-fail  
Always       -       0
   7 Seek_Error_Rate         0x002e   200   200   000    Old_age   
Always       -       0
196 Reallocated_Event_Count 0x0032   200   200   000    Old_age   
Always       -       0
197 Current_Pending_Sector  0x0032   200   200   000    Old_age
(Continue reading)

Permalink | Reply |
headers
Christian Franke | 18 Jun 2012 19:17
Picon
Favicon

Re: pending blocks not reliably found in -t long

mathog wrote:
> On 17-Jun-2012 22:38, Christian Franke wrote:
>> mathog wrote:
>
>>> picked up some pending sectors.  Strangely these are not always found
>>> with a
>>> smartctl -t long /dev/sda
>>
>> Recently I had a similar issue with a SAMSUNG HE103UJ: 40 pending
>> sectors were reported but both SMART long test and a read scan with
>> ddrescue didn't report any bad sectors. The pending sector count was
>> reset after zeroing each disk block with ddrescue.
>>
>> Apparently this firmware resets the pending status only on write even
>> if the read error was transient. This is vendor specific and typically
>> never documented.
>
> How did you find the bad blocks if subsequent reads completed?

Never :-)

>
> Also bizarre on the WD10EARS drive was what happened when the bad 
> blocks were overwritten, this is the final state:
>
>   5 Reallocated_Sector_Ct   0x0033   200   200   140    Pre-fail  
> Always       -       0
>   7 Seek_Error_Rate         0x002e   200   200   000    Old_age   
> Always       -       0
> 196 Reallocated_Event_Count 0x0032   200   200   000    Old_age
(Continue reading)

Permalink | Reply |

Gmane