I found out the answer. I did some testing with files and /dev/loop to emulate the same scenario. The answer is No. Data from 2 drives of a raid-5 cannot be mixed. ddrescuing one failed drive over the other failed drive, has no meaning at all. The data stored in each drive are only valid if the assembled drive get its proper position in the array. Thanks Costa 2013/8/30 Tsaousis, Costa <costa@xxxxxxxxxxx>: > Hi all, > > I have a raid5 with 4 drives: sda, sdb, sdc, sdd. > > While I was in vacations sdb powered off (I guess a thermal issue). Anyway, > the raid continued for several days in decreaded mode. > > One day before being able to replace the powered off drive, sdd, a second > drive, broke (lots of read errors, including the raid superblock). > > sdb (the first drive) works, but it is several days behind the other 3. > > I thought the best way to get the most out of a possible reconstrution, is > to ddrescue as much of sdd to sdb the then attempt to re-assemble the array > with the 3 drives (2 fully functional + sdb with as much of sdd's data as > possible), hopping that if I am lucky, the parts that cannot be read from > sdd, will exist in sdb and may not have been changed in the few days that > sdb was offline. > > Is this going to work? Can I just "mix" data from 2 raid-5 drives to build > an operational copy of the failed drives? > > Thanks. > > Costa > > PS: I have backup of important data, and I already have restored > operations. However the array was also hosting a few TB of > nice-to-have data, never backed up, which I am trying to rescue if > possible. -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
