Thread (5 messages) 5 messages, 3 authors, 2009-09-18

Re: Raid 5 Issue, cannot recognize EXT3 File system.

From: Robin Hill <hidden>
Date: 2009-09-18 08:13:13

On Thu Sep 17, 2009 at 05:46:33PM -0400, Sunpyo Hong wrote:
First off, lemme tell you the initial problem. I had a WD ShareSpace that
had one of the disk go bad. They sent a replacement and it was suppose to
rebuild on its own, however after the build, the array went bad and it was
no longer able to see any of the files. 

I downloaded and tested the drives using windows data recovery tools I saw
that the ext3 was Linux FS and that using these tools would not help in the
recovery. However through the tools I was able to see and recover some of
the files, but the files themselves were usable. I confirmed with WD that
ext3 was in fact the FS and took steps to recover the data. These are the
steps I took in order for me to assemble the raid.

Right now I have 3/4 drives with the data. I did #mdadm --assemble --scan,
which let me assemble the raid. However at this point I was not able to see
any of the files or mount the drive to the mount point it was once at. I
have also tried #mdadm --create with the array in the right order /w the
missing disk. 

Initially the --assemble --scan assembled the array /dev/md2 with the disks
in the wrong order. I know because I physically saw where the disks were in
relation to the disk order and wrote down the disk order on every HD.
This should not be possible - --assemble will read the metadata from the
superblocks and determine the order the drives should be within the
array.  The physical positioning/ordering is irrelevant here.  I suggest
you recreate the array in the original order --assemble did (and hope
that the chunk size used is the default).
#cat /proc/mdstat
Personalities : [raid1] [raid6] [raid5] [raid4] 
md2 : active raid5 sdb4[1] sdd4[3] sdc4[2]
      5856150144 blocks level 5, 64k chunk, algorithm 2 [4/3] [_UUU]
      
md0 : active raid1 sdd1[0] sdb1[2] sdc1[1]
      208768 blocks [4/3] [UUU_]
      
I find it somewhat odd that md0 and md2 have the drives in a different
order.  Is this related to your recreating the array?  Admittedly, md0
is only a RAID1 so the order doesn't actually matter, but if the arrays
were created at the same time then the normal behaviour would be to use
the same order in the create command.

Cheers,
    Robin
-- 
     ___        
    ( ' }     |       Robin Hill        [off-list ref] |
   / / )      | Little Jim says ....                            |
  // !!       |      "He fallen in de water !!"                 |

Attachments

  • (unnamed) [application/pgp-signature] 198 bytes
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help