• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 774
  • Last Modified:

Windows Server 2008 - Recovery from Raid 0 Disk Replacement

I have a Dell Server with a 4-disk Raid 0 configuration on a SAS 6/iR controller.  Physical Disk 0 is predicating failure and I have a replacement disk already in hand from Dell.  My problem is how do I backup the data before I delete the virtual disk and replace the physical drive so I can then do a restore/recovery?  AppAssure will not backup the drive because of physical disk problems, and Windows Server Backup fails for the same reasons.  Can I just do a Windows Server Backup and Restore/Recovery of the operating system and backup/restore the other shared volumes through another method?  What about imaging the server - any thing that would be fast and cheap to get this done?
0
kdlange77
Asked:
kdlange77
  • 5
  • 5
  • 2
  • +2
1 Solution
 
xDUCKxCommented:
RAID 0 or RAID 1?  RAID 0 (Striping) if you take the drive out your computer won't come back up.  You'll need to reimage.

Snap Deploy can do an image of your machine:

http://www.acronis.com/enterprise/products/snapdeploy/

If it's RAID 1 (Mirroring)  you should be fine.  The array will rebuild in real time and you should be able to maintain operation with a little bit of degredation in performance.
0
 
kdlange77Author Commented:
It is Raid 0...and I already know that I must backup and restore after deleting and creating the virtual disk.  I will check out Acronis Snap Deploy.  Thanks...will let you know if that works.
0
 
jason_0573Commented:
Why on earth are you using a RAID 0 for 4 disks?
0
Free learning courses: Active Directory Deep Dive

Get a firm grasp on your IT environment when you learn Active Directory best practices with Veeam! Watch all, or choose any amount, of this three-part webinar series to improve your skills. From the basics to virtualization and backup, we got you covered.

 
kdlange77Author Commented:
Pardon me, but that is not helpful.  Why on earth are you here?
0
 
DavidCommented:
Do a windows chkdsk with the option to repair & replace bad blocks.   This MAY fix the problem, depending on whether it is a file, swap, free area, etc..  Perhaps the bad block(s) are not part of a file you care about, like swap space.

If that doesn't do the trick, one has to identify the bad block(s) at the logical level and force a write to them. If that area is free, no problem.  If it is part of a file or something important, you lose data and don't know which files are affected (but that too can be resolved by writing some code and/or looking at logs, or even just reading all files into the bit bucket)
0
 
kdlange77Author Commented:
I have already run chkdsk numerous times and repaired some bad blocks, but the problem remained AND Dell has already determined the disk needs to be replaced and has sent it to me and I am in the process of replacing it.  Thanks.
0
 
DavidCommented:
Here is an easy thing to do if you want to just fix the problem and move on (recognize risk of destroying a file w/o knowing the name of it)

1. Boot to linux via a usb drive Use a decent "server" distribution that has the native drivers (Like centos).
2. Assuming the O/S detects /dev/sdb (and /dev/sdb1, .. n), where these represent partitions ..

3.  dd if=/dev/sdb of=/dev/null bs=64k   (This is a safe read only into the bitbucket, nothing gets written, if it blows up with error at block xxxx, then you have confirmed uncorrectable read error.  

4. Repeat above, using bs=512 for 512 bytes at once, use the start= option to start at a point before the error.   this will tell you which block is unreadable.
5.  Then use dd if=/dev/zero of=/dev/sdb  with the start/count variables to issue a controlled write.  That overwrites.

DANGER DANGER .. screw this up, and you could easily blow away everything.  So I suggest testing the process on a scratch USB drive.  look at man pages of dd and make 100% sure you understand the risk.  

There is no are-you-sure in linux.  Enter wrong block, or device, or whatever, and the computer will gladly blow away your O/S.
0
 
kdlange77Author Commented:
Without trying to sound unappreciative...what the F... are you talking about?

I am not trying to repair the disk...it needs to go away.  I just need to backup my server someplace so I can restore it after I replace the drive.  I am pretty sure Acronis Snap Depoly will work...so please...no other suggestions or helpful comments until I try this and see if it works.
0
 
DavidCommented:
If you manually overwrite blocks,  say block # 12345-12347, you could do this first .

dd if=/dev/sdb of=/tmp/scratch.bin start=12340 count=5

That gives file /tmp/scratch.bin  Perhaps the contents of /tmp/scratch.bin will reveal a clue as to the file it is part of.  

Now  block #12345 isn't necessarily part of the same file that the blocks immediately before it was, but in order to answer that question, you need to get the checkbook out and hire a pro (no I am not fishin'g for a gig), just letting you know that there is no simple easy command to cross reference such things, so you have to pay somebody who has written such code or can determine it by taking apart file systems.
0
 
DavidCommented:
You can NOT make an unrecoverable read error go away w/o writing that chunk of data (or making that chunk of data unnecessary).  You can not write to any arbitrary chunk of data with any built-in windows program.  

So either use a LINUX shell script so you don't have to write a program, or write a C program that talks to the hardware directly with unbuffered I/O and do it yourself.
0
 
DavidCommented:
Your software is going to continue to halt reading that particular physical block (where there is one there is often many), until that block has been overwritten.  

Since chkdsk failed and backups fail, it is a safe bet that this particular chunk isn't tied to a file as much as it is tied to some supporting data, like a file descriptor.
0
 
noxchoCommented:
If AppAssure fails to read the bad blocks then you need simply try different software that can pass the bad blocks and continue with those which are healthy.
And if you run CHKDSK /r already then Windows marked them as bad and decent software should be aware of this fact as well.
Try Drive Image or R-Studio http://www.drive-image.com/ 
They both are capable of creating backup image of a bad drive (I used R-Studio mostly).
Also give a try to Paragon Drive Backup 11 Server trial: www.drive-backup.com
0
 
xDUCKxCommented:
Wow...he's not trying to do forensic analysis on the drives.  He just wants a backup.  Try acronis, or if you have a Virtual Server (VmWare, Hyper-V) you could even try cloning it.  VMWare Converter should be able to image the drive while the server is running.  Acronis should be quick and easy though.  You'll need an external USB Drive to dump the image(s) to.
0
 
kdlange77Author Commented:
I have already contacted Acronis and downloaded a trial version.  It will be able to bypass the bad blocks, and I have identified the files that are corrupt already and they are not critical files and can be replaced by other means.  I am going to close this questions and award the points to the first responder who pointed me to Acronis.

Thanks xDUCKx for understanding my intent and pointing me in the right direction.
0

Featured Post

Hire Technology Freelancers with Gigs

Work with freelancers specializing in everything from database administration to programming, who have proven themselves as experts in their field. Hire the best, collaborate easily, pay securely, and get projects done right.

  • 5
  • 5
  • 2
  • +2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now