Re: "Repair" hard disk with bad sectors by reading and rewriting same data?

"Percival P. Cassidy" <[email protected]> wrote:

> On 8/15/20 12:24 AM, VanguardLH wrote:
>> "Percival P. Cassidy" <[email protected]> wrote:
>> 
>>> On 8/14/20 4:44 PM, VanguardLH wrote:
>>>> VanguardLH <[email protected]> wrote:
>>>>
>>>>> Seagate has their own SeaTools you might try.  Make a bootable CD from
>>>>> which to run their software.  That way, the bad HDD is quiescent, and it
>>>>> doesn't matter what OS is there.  Any remapping is done at the hardware
>>>>> level, not at the OS level in whatever file system the OS uses.
>>>>>
>>>>> https://www.seagate.com/support/downloads/seatools/
>>>>> Get the bootable version.  Then you don't have to install it in any OS,
>>>>> and you don't have to move the HDD between computers.
>>>>
>>>> I found:
>>>>
>>>> https://www.hdsentinel.com/usbboot.php
>>>>
>>>> I'm guessing it uses an image of their free DOS edition mentioned at:
>>>>
>>>> https://www.hdsentinel.com/hard_disk_sentinel_dos.php
>>>>
>>>> I still don't see in its description that it does any testing to prod
>>>> the HHD's firmware to remap the bad sectors.  Seems to be just for
>>>> analysis.  The Seatools might just be a diagnostic tools, too.
>>>>
>>>> I'm surprised smartctl didn't get remapped the bad sectors/blocks.  Bad
>>>> blocks is one of its functions.  If there are no spare sectors reserved
>>>> for remapping, then there's no way to remap any more bad sectors.
>>>>
>>>> There is no information in SMART or by reading an HDD's firmware
>>>> signature to know the size of the pdata table.  The p-list table is the
>>>> remapping done at the factory.  g-list (growth list) is used for further
>>>> remapping during the use of the drive.  I've seen where there was a jump
>>>> in bad sectors that got remapped (pending count went up, and then fell
>>>> to zero), and it could be a couple hundred sectors; however, without
>>>> knowing the total size of the g-list table (to know how many spare
>>>> sectors are available), there is no way to know how much is left.  The
>>>> drive makers don't publish the size of their g-list table.  200 bad
>>>> sectors is really bad if there are only 400 spare sectors total for
>>>> remapping, but not immediately serious if there are 20,000 spare
>>>> sectors.  Bad sectors tend to grow radially on spinning magnetic media,
>>>> so if some show up then it becomes more likely that more will show up.
>>>> You cannot know severity of a remap event when the total or remaining
>>>> capacity isn't known.
>>>>
>>>> But if the HDD's firmware doesn't remap the bad sectors, that could be
>>>> because there are no more spare sectors.
>>>>
>>>> https://dtidatarecovery.com/how-to-fix-bad-sectors-on-a-hard-drive/
>>>> "Since we know that all bad sectors are remapped to a pool of unused
>>>> sectors, and the size of this pool is substantial, the only way a bad
>>>> sector will show up on your system is if the pool has been completely
>>>> used. In other words, your remapping pool is so full that it cannot take
>>>> another bad sector and now the drive is throwing errors that it can’t
>>>> read from a sector."
>>>>
>>>> Start checking on prices for replacement HDDs of equal, or greater,
>>>> capacity to which you can clone the old HDD.  If possible, see if you
>>>> can determine in which files are the bad sectors.  The clone software
>>>> should keep a long, too, to let you know if there were read errors from
>>>> the old drive.  If they are data files, you can recover those from your
>>>> backups.  If they are program files, well, you could uninstall and
>>>> reinstall the programs to ensure all their bits are correct.  If the bad
>>>> sectors are in system/OS files (which rarely get rewritten except by
>>>> updates, and then only some of the OS files get replaced), well, you'll
>>>> have to see if the cloned HDD is bootable, the OS behaves, and you
>>>> experience no problems, or reinstall the OS on the new HDD, reinstall
>>>> all the apps, and restore the data files to it.  If your backups include
>>>> full-partition images, you can obviously not bother with the clone
>>>> operating and just restore the backup image to the new drive.
>>>>
>>>> If the pending count doesn't go down, you have a drive that won't heal.
>>>
>>> The Pending Count alwsys does go to zero after writing zeroes to the
>>> whole drive.
>> 
>> A bad sector does not become a good sector because of what bit string
>> was written to it.  That's like saying a crack plate atop of which is
>> chocolate cake becomes uncracked because the chocolate cake got removed
>> to put angel food cake atop the plate.
>> 
>> Just how are you determining there are bad sectors?  More likely what
>> happened is that those bad sectors where not in use by files that were
>> getting rewritten.  The write operation (1's or 0's) will generate the
>> sector error whereupon the drive's firmware will remap the bad sectors
>> to spare/reserve sectors.  Something has to use the bad sector for the
>> drive to see there are errors using the bad sector and to remap it.
> 
> I write zeroes to the whole drive, then run a long smartctl test, 
> followed my smartctl -x. Pending and Uncorrectable counts are then found 
> to be zero but there is no indication that any sectors have been remapped.
> 
> This is using smartctl in FreeNAS (based on FreeBSD).
> 
> Perce

Understood.  I didn't realize you were willing to destroy all files
currently on the drive to resolve the bad sectors getting remapped.