[ale] Strange drive (fdisk) output?

[Michael B. Trausch]

On Mon, 2007-01-29 at 23:35 -0500, Christopher Bergeron wrote:

> Thanks for the tips!   Is there a way I could take a bootsector from the 
> working drive and copy it onto the failed drive?
> Perhaps:
> dd if=/dev/hda of=/dev/hdb bs=1024?  What bs size should I specify, and 
> does it even matter?
> 
> In fact, now that I think about it, I may have killed the drive that 
> way.  I was mirroring my drives using dd, so maybe I specified an 
> incorrect block size and corrupted the drive?
> 
> Is there any way to correct the geometry of the drive using utilities, 
> or would my best bet be to copy the boot sector from another drive?  
> Failing that, I'm going to look into the warranty items (assuming, I 
> didn't corrupt it myself).
> 
> Kind regards and thanks again!
> Chris Bergeron


I am not sure if this is still the case or not, but I know that once
upon a time I had a shiny new 4 gig drive that was under warranty.  I
used it to transport data between home and school, and so it was in one
of those 5?" housings that you could take from one computer to the
other.  One day, on my way into school, I tripped and the drive flew
from my hands.  The plastic bay shattered, and the drive bounced.  And
bounced.  And bounced again.  Several more times, in fact.  WD still
replaced it as a warranty RMA.  They don't usually ask questions.

That having been said, I don't know if they're still that lenient.  I
haven't physically damaged a drive in quite a long time.

If the data on the drive is important, attempt to recover it before
doing anything to it.  Then, the best way that I have found to reset
drives that are messed up is to 'dd if=/dev/zero of=/dev/hdb bs=512
count=1M' on them.  After the dd is finished, you'll have one million
blocks of nothingness.  You can then use Linux fdisk or cfdisk to
generate a new MBR and partition table for the drive, which it will
quite happily do if the drive is working.  If you suspect that the
blocks on the drive might be going bad and that the bad block reserve
has been diminished, you can omit the count parameter, and it will write
zeros to the entire disk, until either (a) it reaches the end of the
disk or (b) it encounters an I/O error.

I was thinking about something, though.  You mentioned that one drive
was 120GiB and that the other drive was 120GB, but that they're both
labeled as being 120GB.  Were they purchased at the same time?  If they
weren't, it's possible that they changed the base system (e.g., from
binary gigabytes to decimal gigabytes) and never bothered to update the
model number.  I have seen that happen at least once before.  If that's
the case, then a binary image of the drive that is 120GiB won't work on
the 120GB drive, and vice versa.  You could clone the filesystems or
something, though.  I tend to prefer transfer solutions that are not
dependent on the block structure of a disk.  dd is quite agnostic about
such things.  :-)

    -- Mike

--
Michael B. Trausch
                    fd0man at gmail.com
Phone: (404) 592-5746
                          Jabber IM:
                    fd0man at gmail.com
              fd0man at livejournal.com
Demand Freedom!  Use open and free protocols, standards, and software!
-------------- next part --------------
An HTML attachment was scrubbed...

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part