I believe you have several mis-conceptions:
 
1. bad-blocks are not created by reading/writing disks. They are either 
manufacturing defects or induced by particles inside the disk, or a vibration 
shock that can scratch the surface while the disk is in operation.

2. A disk with bad blocks can be used - it needs a low level format -
a process that includes surface tests, and in which the disk controller 
itself re-assignes spare blocks instead of bad blocks. Of course there's 
a limit to thos spare blocks, so the suceess really depends on the amout
of damage you have in your disk.

3. Applications don't deal with bad clocks (not even bittorrent) 
file-systems may or may not have a level of bad blocks management (on top of
the disk controller). Typically this requires the FS to do a write-verify, but
most day-to-day FS don't do that.

Use the bad disk on a windows system - run a low level format on it including 
full blown surface test (its much harder to do the same on linux). Then
use the disk as a good disk and pray a lot.

Good luck,
Dan


> -----Original Message-----
> From: [EMAIL PROTECTED] 
> [mailto:[EMAIL PROTECTED] On Behalf Of Beni Cherniavsky
> Sent: Sunday, September 03, 2006 1:22 AM
> To: linux-il
> Subject: Utilizing a dying HD for unreliable storage
> 
> I have an HD that recently got some bad sectors (*).
> I suspect it was facilitated by lots of bittorrent work and I'm not
> sure I want to subject my new HD to the same fate.  And anyway I don't
> want to throw the old away if I can use it.
> So I thought: bittorrent should be able to cope with any corruption in
> file contents.  Let it work on the old HD.  However, directories and
> other FS structures must reside on the new HD, to avoid major
> corruption from few lost sectors.
> 
> Is there a filesystem that can store all metadata on a different
> device from the file contents, or can be easily mutated into doing so?
> I found reiserfs can store the journal on a different device but I
> don't think that's enough.
> I could write something quick&dirty with FUSE & Python but it won't
> have good performance & bug-freedom without a lot of work...
> 
> (*) in case anybody cares for the story:
> - Beat me on the head for burning movies every week but not a single
> backup in years!
> - Beat me on the head for not enabling smartd.
> - Install debian on new HD
> - GNU ddrescue recoved to an image all but 430KB of the 80GB disk :-)
> - Still heavy fs corruption (unlucky bad sectors), needed reiserfsck
> --rebuildtree
> - 3700 files in lost+found, several files deleted
> - Practically no damage to /home, /etc, /usr/local :-)  apt-get will
> fix the rest
> 
> -- 
> Beni Cherniavsky <[EMAIL PROTECTED]>, who can only read email 
> on weekends.
> 
> =================================================================
> To unsubscribe, send mail to [EMAIL PROTECTED] with
> the word "unsubscribe" in the message body, e.g., run the command
> echo unsubscribe | mail [EMAIL PROTECTED]
> 
> 

Reply via email to