help with recover data from software raid 1

[Franz Waldmüller]

Hi Luis,

Am 2011-08-02 03:59, schrieb Luis Paulo:
> 2011/7/31 Franz Waldmüller<waldbauernbub at gmx.at>:
>> Hi,
>>
>> I want to recover files from a disk which has been part of a raid1 array. I
>> don't want to mount it degraded, as all operations must be read only.
>>
>> Can I mount the partiton of the disk as an ext4 filesystem (the raid was
>> formatted as ext4.)?
>>
>> Background:
>> In my server to harddisks are used in mirrored array to save all important
>> data.
>> /dev/sdb1 and /dev/sdc1
>>
>> today I created another md-device with the created partitions /dev/sdb2 and
>> /dev/sdc2. Unfortunately I didn't edit the /etc/mdadm/mdadm.conf file
>> properly.
>> When I rebooted the server I checked /proc/mdstat
>> and saw the following resync process:
>>
>> md1 : active raid1 sdc[1] sdb2[0]
>>       391382976 blocks [2/2] [UU]
>>       [>....................]  resync =  0.1% (745088/391382976)
>>
>> That is wrong, as the first device is sdc, but it should be sdc2.
>> I halted the server and now have to recover the files from /dev/sdb1.
>> /dev/sdc1 could have suffered damage from the resync process.
>>
>> I hope that you can help me with your advice.
>>
>> Franz
>
>
> I don't understand why you don't want to assemble the array as degraded.
>
> The way I'll recover the array will be clearing the super-block of
> /dev/sdc1, assemble the array as degraded with dev/sdb1 and then add
> /dev/sdc1 (using mdadm commands). At least this is the general idea, I
> may (and you must before trying it) look more carefully into it.
> After a resync, the array would be functional again.
>
> Regards
> Luis
>

Thank you for your advice.
I am not sure about two things:
Does this output from mdadm mean, that data is read from sdc and written 
to sdb2?
md1 : active raid1 sdc[1] sdb2[0]
        391382976 blocks [2/2] [UU]
        [>....................]  resync =  0.1% (745088/391382976)
In this case everything would be fine, as sdb2 was empty. If it is the 
other way round things are more severe as it would mean writing zeros 
from sdb2 to sdc.

The other option with mounting as degraded is good, too but I as soon as 
I noticed this strange resync process I shut the server down 
immediately. It was the only thing I could think at this moment.

The harddisk are connect by usb and up to now I don't know how to find 
out which disk is sdb and which sdc? I attached the disk now to another 
workstation (backup all data). But is there a possibility look the 
serial number of the drives up in the server logs?

Thanks for your help
Franz