[SLL] get md RAID array to rebuild even if read errors?

[Jeremy C. Reed]

On Thu, 14 Feb 2008, Rob Smith wrote:

> On Thu, Feb 14, 2008 at 12:44 PM, Jeremy C. Reed <reed at reedmedia.net> wrote:
> > Any ideas on how to get mdadm or md to rebuild even if read errors?
> 
> --force doesn't do it?

Sorry I should have mentioned that. My last test was to restart the sync 
with:

  /sbin/mdadm --force --add /dev/md0 /dev/sdc2

It failed a few minutes later same way -- and now is starting at 0.0% 
again.

> Another option is to use dd, dd from one to the other and it'll
> pretend it's the correct drive to mdadm

I may try that next after I stop (remove) the second disk from the array 
and after a complete backup to a third disk (not part of raid). I guess 
the dd will continue on the read error and at the area will just copy -- 
maybe -- corrupted data.

Using dd to copy will probably be easier and faster than: creating a new 
raid on second disk; creating file system; copying all data; and getting 
system to boot from second raid.

Thanks,



  Jeremy C. Reed