Moving LVM data off a bad disk

Moving LVM data off a bad disk

[Daniel Ahlberg]

Hi,

Recently one of my harddisks began to develop bad sectors. I decided to add a 
another disk to replace the disk and run pvmove -i to salvgage as much as 
possible to the spare disk. But, the process takes seems to take ages. For 
every bad sector pvmove encounters it takes 5 seconds for the kernel to 
timeout and move on to the next sector. Some calculation would tell me that 
it would take approx 6 years to go through the rest of the disk.

So, is there any way to speed up this? I was thinking in the ways of reducing 
the 5 second timeout or having the disk driver simply skip the bad sectors.

These are the errors I get:

Aug  4 00:01:31 springbird kernel: hdc: dma_intr: status=0x51 { DriveReady 
SeekComplete Error }
Aug  4 00:01:31 springbird kernel: hdc: dma_intr: error=0x40 { 
UncorrectableError }, LBAsect=34880281, sector=34880218
Aug  4 00:01:31 springbird kernel: end_request: I/O error, dev 16:01 (hdc), 
sector 34880218
Aug  4 00:01:36 springbird kernel: hdc: dma_intr: status=0x51 { DriveReady 
SeekComplete Error }
Aug  4 00:01:36 springbird kernel: hdc: dma_intr: error=0x40 { 
UncorrectableError }, LBAsect=34880283, sector=34880220
Aug  4 00:01:36 springbird kernel: end_request: I/O error, dev 16:01 (hdc), 
sector 34880220