icheck and ncheck on ReiserFS to convert bad sector numbers to filenames

icheck and ncheck on ReiserFS to convert bad sector numbers to filenames

[Chris Witham]

I have a hard drive which has failed.  I posted about it here[1], and
was referred to ddrescue[2] to create an image of the failed disk.
This was successful, and I now have an image of the drive, including
the 1688 bad sectors that couldn't be read.  The logfile lists all the
the bad sectors, and I was referred to these[3] instructions to
convert the bad sector numbers into filenames that are corrupted.

Unfortunately, those instructions involve debugfs, which is
ext2/ext3-specific.  What little information I was able to find online
about the icheck and ncheck commands used in the instructions and
ReiserFS has led me to either complaints that they are not implemented
for ReiserFS, or some variant of a post[4] to this mailing list that
suggests using 'find / -type f -exec cat {} > /dev/null \;' at
watching for the filenames that fail.

That seems like a complete brute force method (scanning over 110GB to
locate 800kB of bad data!), and will almost certainly return
inaccurate results because the drive is continuing to fail and will
return more problems than it experienced initially when the drive
image was made.  (Not to mention stress the disk further and make any
more work I may need to do on it more difficult.)

Is there any way to do the equivalent of icheck or ncheck in ReiserFS?
 If not now, are there plans to implement this functionality anytime
soon?


[1] http://www.linuxquestions.org/questions/showthread.php?t=485322
[2] http://savannah.gnu.org/projects/ddrescue/
[3] http://smartmontools.sourceforge.net/BadBlockHowTo.txt
[4] http://marc.theaimsgroup.com/?l=reiserfs&m=114425109321290&w=2

Re: icheck and ncheck on ReiserFS to convert bad sector numbers to filenames

[Edward Shishkin]

Chris Witham wrote:

> I have a hard drive which has failed.  I posted about it here[1], and
> was referred to ddrescue[2] to create an image of the failed disk.
> This was successful, and I now have an image of the drive, including
> the 1688 bad sectors that couldn't be read.  The logfile lists all the
> the bad sectors, and I was referred to these[3] instructions to
> convert the bad sector numbers into filenames that are corrupted.
>

Make sure that blocksize for copy operation performed by dd_rescue
was specified as Reiserfs blocksize (4096) and bad block images were
zeroed, then run reiserfsck on the backup.

> Unfortunately, those instructions involve debugfs, which is
> ext2/ext3-specific.  What little information I was able to find online
> about the icheck and ncheck commands used in the instructions and
> ReiserFS has led me to either complaints that they are not implemented
> for ReiserFS, or some variant of a post[4] to this mailing list that
> suggests using 'find / -type f -exec cat {} > /dev/null \;' at
> watching for the filenames that fail.
>
> That seems like a complete brute force method (scanning over 110GB to
> locate 800kB of bad data!), and will almost certainly return
> inaccurate results because the drive is continuing to fail and will
> return more problems than it experienced initially when the drive
> image was made.  (Not to mention stress the disk further and make any
> more work I may need to do on it more difficult.)
>
> Is there any way to do the equivalent of icheck or ncheck in ReiserFS?
> If not now, are there plans to implement this functionality anytime
> soon?
>
>
> [1] http://www.linuxquestions.org/questions/showthread.php?t=485322
> [2] http://savannah.gnu.org/projects/ddrescue/
> [3] http://smartmontools.sourceforge.net/BadBlockHowTo.txt
> [4] http://marc.theaimsgroup.com/?l=reiserfs&m=114425109321290&w=2
>
>

Re: icheck and ncheck on ReiserFS to convert bad sector numbers to filenames

[Jeff Mahoney]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Chris Witham wrote:
> I have a hard drive which has failed.  I posted about it here[1], and
> was referred to ddrescue[2] to create an image of the failed disk.
> This was successful, and I now have an image of the drive, including
> the 1688 bad sectors that couldn't be read.  The logfile lists all the
> the bad sectors, and I was referred to these[3] instructions to
> convert the bad sector numbers into filenames that are corrupted.
> 
> Unfortunately, those instructions involve debugfs, which is
> ext2/ext3-specific.  What little information I was able to find online
> about the icheck and ncheck commands used in the instructions and
> ReiserFS has led me to either complaints that they are not implemented
> for ReiserFS, or some variant of a post[4] to this mailing list that
> suggests using 'find / -type f -exec cat {} > /dev/null \;' at
> watching for the filenames that fail.
> 
> That seems like a complete brute force method (scanning over 110GB to
> locate 800kB of bad data!), and will almost certainly return
> inaccurate results because the drive is continuing to fail and will
> return more problems than it experienced initially when the drive
> image was made.  (Not to mention stress the disk further and make any
> more work I may need to do on it more difficult.)
> 
> Is there any way to do the equivalent of icheck or ncheck in ReiserFS?
> If not now, are there plans to implement this functionality anytime
> soon?

I'm afraid the icheck/ncheck way of doing things isn't much less
brute-force, it's just hidden from you. You could do what you're looking
for with a perl script that parses the debugreiserfs output.

Otherwise, I'm afraid I don't think there are any plans to implement
what you're asking for.  We accept patches. :)

- -Jeff

- --
Jeff Mahoney
SUSE Labs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org

iD8DBQFFHHCZLPWxlyuTD7IRAj8OAJ0TWJL8FHept81jKTZljL4LYR++IQCglvaL
XVqC4ljJme98HuO4rmtD250=
=fgDc
-----END PGP SIGNATURE-----

Re: icheck and ncheck on ReiserFS to convert bad sector numbers to filenames

[Chris Witham]

On 28/09/06, Jeff Mahoney <jeffm@suse.com> wrote:
>
> I'm afraid the icheck/ncheck way of doing things isn't much less
> brute-force, it's just hidden from you. You could do what you're looking
> for with a perl script that parses the debugreiserfs output.

 I ended up using the brute-force 'find' to scan the failing HD, and
it only identified 3 unimportant files as corrupt.  I restored the
files from the disk image, deleted the 3 bad files, and everything
seems to be OK from there.

 Thanks for the help!


Chris