Server Specs & Partition Setup

• Raid 10
• 4- Western Digital WD5001ABYS-18NA0
• Windows 2003 R2 Server
• Partition 1: Dell Partition. 70.6MB
• Partition 2 (Boot): OS. 12GB, 2.66GB free
• Partition 3 (Data): 918.4GB, 350GB free

Crashed Dell PowerVault 600 Server: Failure Analysis

1. Stripe 0: Hard Drive 0 & 1
   • Disk-0 & Disk-1are good (Verified to Western Digital specifications with no bad sectors)
2. Stripe 1: Hard Drive 2 & 3
   • Disk-2: Failed completely, Head Crash
   • Disk-3: S.M.A.R.T. reports Immanent Failure Very slow with many bad sectors
3. Dell Raid controller failed to recognize the S.M.A.R.T. status of Hard Drive 3: "Immanent Failure"
4. Client's IT Support attempts to save the data:
   • Server still boots very slowly
   • Backup of the Data Partition Failed (partial copy only)
   • Dell Support sent a new drive
   • Attempt to rebuild the Raid failed at 15%
   • Server then failed to boot
   • Dell Support suggests to re-initialize the Raid Array and re-load the OS
   • Client decided to quit and bring us the Dell Server for Data Recovery

Raid Data Recovery: procedures

1. Duplicate as much of each hard drive of the Server immediately
   • We use our own in-house software to duplicate drives with bad sectors
2. Determine the order and stripe size of the Raid Array
   

   You can determine the stripe size by looking at the $MFT record structure to determine the hard drive order (click to enlarge) The File Record Number of the first record is 00 00, followed by 01 00... 02 00...etc. Scroll down the next sectors, you will find the sequence of the Record Number will jump. Keep on following the sequence until the next jump, you should find the Stripe pattern of 128 sectors for each jump. The next 128 sector sequence will be on the second drive of the Stripe Set at the same starting sector. This applies to the standard 64K (128 sector) Stripe of the Dell PowerVault 600 default Raid settings.

   
   

   Once the Drive Order is known, find the 1st drive in the Stripe. This contains the Partition Table Sector (click to enlarge)

   
   
3. A duplicate of the Disk-3 mirror was made and replaced the original failed Disk-2 artificially rebuilding the Raid 10.
4. Verify the Raid Configuration in the Bios settings <CTRL>-R. Since the settings were incorrectly modified by the user, we created a new virtual drive with the correct settings for Raid 10 being sure NOT to Initialize the drives. This just forces all the drives back online.
5. The Server then booted up with no problem and all the data was intact.