From: Brad <bra...@gm...> - 2009-06-29 22:27:45
|
2009/6/30 Bruno Wolff III <br...@wo...>: > On Mon, Jun 29, 2009 at 16:44:29 +0200, > Frédéric Boiteux <fbo...@ca...> wrote: >> >> Try to write a number of sectors used for one block of the filesystem >> which uses the bad sector : often, writing one sector isn't >> sufficient (you'll have to verify that all these sectors aren't used >> by the filesystem before, as explained in Bad Block Howto). > > Possibly what is happening is that because he is only writing a partial > block, the OS is first trying to read the the original block so that it > can preserve the parts that won't be changing. When this operation fails, > it blocks the write that would trigger reallocation of the bad sector. > Writing using the OS blocksize (typically 4096 on linux systems) properly > aligned should work around that issue. Well, I would have thought that the 'dd' commands were explicit and direct enough, targeting just that specific sector, and bypassing any filesystem or OS massaging. However, I'll spend a few hours saving its contents and then aim a 'dd' command with an 8K block size at the area all around the bad block. :-) The disk is in an external caddy (with a SATA 2 connection) and it's formatted with an NTFS filesystem so I wouldn't know how to track down the affected files (other than by attempting reads of each one); I'll just backup what I need and then cheerfully blast the immediate neighborhood of that bad sector. If all this fails what would be my next step? A low-level disk format? Are there tools that do that under Linux, or is it a case of finding something on the Western Digital web site? Thanks, Brad |