[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On 16/07/12 09:38, George Parker wrote: > > Simon, there is nowhere I have found that says DD swaps > out bad blocks when you write. It isn't "dd" that does it. The drives firmware will replace with a "spare" any blocks that a write fails on. The intent is that on first usage any bad blocks are set aside. "dd" is just the most convenient method of writing to every block on a disk to trigger this behaviour. I don't trust the "dd" method. Indeed I replace any disk that encounters new bad blocks in normal operation, based on both the underlying engineering and experience of trying not to. Also the "blocks" are so small on disk, it is rare to encounter a single new bad block these days, they inevitably comes in their tens or more usually hundreds, no doubt reflecting the track of some minute speck of dust or scratch. In theory with RAID-1 or similar most places we ought to be able to live with a less than perfect disk, simply having it drop out of the RAID array when a read error occurs, in practice I've never seen a disk with a new bad block formatted and reintroduced to a RAID array last out the month. -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq