Hardware RAID 5 with 2 drive failures

[Archaniel]

Elloz,

I know this might be impossible to do, but I'm just looking for options. I have had a double drive failure on an older IBM x3550 M3 server with RAID 5 configured on an LSI card. I didn't really look closer into the state - whether the failed drives spin up yet or if the drive would show up within a linux box - though my idea is `dd`-ing the whole drive to a same (I have few spares) and putting it back into the array - might that work?
I have seen also a 3rd drive in a `predicted failure` state so ... happy days :D

The VDisk had a VMware boot drive + datastore on it and I need to pickpocket an sql database within a centos vhost from there, which I sadly didn't have backed up :/

do you have any other ideas maybe? I'm more or less convinced the data is lost, but I just want to play around with this...

 

[Nick2253]

It really depends on the failure reason.

In theory, if the only problems was a controller board failure on the HDD, it's possible that you could swap controllers on the drives. This is an advanced operation, and may not work unless the drives are exactly the same (same revision and everything).

If you can dd the drive, it's possible in theory that you could simply replace the "failed" drive, but I would wager dollars to donuts that such a operation wouldn't work; if you can pull the data off the drive that simply, the RAID controller should be able to too.

What happens in professional data reconstruction is that the drives are all copied out (often times in clean rooms with custom hardware), and the RAID array data is rebuilt "manually" using the disk images using software. I'm not sure if there is any open source or free software that does this.

A classic trick for getting data from hard drives is to freeze them prior to pulling data off them. You might be able to use this trick to copy the hard drive, and see if that works.

 

[Bidule0hm]

I recommend you to do a dd of each drive before attempting anything so you have an exact backup of the drive you can copy any number of times you want to do your tests/recovery.

In theory, if the only problems was a controller board failure on the HDD, it's possible that you could swap controllers on the drives. This is an advanced operation, and may not work unless the drives are exactly the same (same revision and everything).

You would also need to swap the ROM with the metadata...

A classic trick for getting data from hard drives is to freeze them prior to pulling data off them. You might be able to use this trick to copy the hard drive, and see if that works.

Only works for a very specific failure, not really useful here IMHO.

 

[Archaniel]

ok, well I try to dd the drives if I have time. So basically that's the only possible thing to do. Sounds like a plan :) Thanks for your inputs