SBS 2003 recovery from NTBackup to new RAID volume and partition following Blue Screen and bad stripe
I had a blue screen event on my SBS 2003 R2 server. The apparent cause is a bad stripe on a RAID 5 volume made from Seagate Barracuda 1TB 7200.11 disks. The volume contains the operating system partition. (I know there are issues with these disks now.)
When the server reboots it is highly unstable and a lot of services won't start (e.g. ISA, Exchange, SQL).
I have full C partition (OS location), system state, Exchange MDBDATA and SQL backup conducted by NTBackup for consecutive days before the blue screen.
The C partition is on a RAID5 volume (CVOL) which has a SMART error and bad stripes. This is what I have done.
1) I have created a new RAID1 volume (DVOL) of the same size as CVOL on new disks attaached to the same RAID controller. It's a RAID1 because I only had 2 free ports on the controller.
2) I booted the server in safe mode and, using server management console, created a new D partition on the DVOL volume. I formatted it.
3) I then rebooted the server in DSRM safe mode and used NTBackup to recover the C partition to the new D partition, and system state from the NTBackup from two days before the blue screen. I chose the option to replace all of the files on the destination (although it was empty of course) and maintain load points (not sure about this).
4) I rebooted the server, entered the RAID array control utility (Ctrl-A as it's an Adaptec array) and promoted DVOL ahead of CVOL in the boot order, restarting the server when the change was accepted.
Problem. The server doesn't boot into SBS 2003 R2. I just get a blinking cursor. F8 to select the boot options such as safe mode doesn't work. I have to power down and reverse point (4) to get an OS load.
Question 1. What am I doing wrong?
Question 2. How do I create a complete restore of the OS partition and system state but to a new Volume on the same hardware, so that I can avoid the SMART error and bad stripes that plague the original Volume?
Question 3. Is there anything specific I need to do so that I can eventually recover Exchange and all my SQL data bases?
Queston 4, alternatively, can I restore/repair to the original C partition on CVOL if I run chkdsk /r by booting the server from disk 1 of the SBS install CD-ROM and chosing to repair an existing installation? (For information, chkdsk /r has been running for 12 hours and has reached 54% on step 4 of 5. There are approximately 100,000 files on the C partition and they ocupy 140GB. CVOL is a 245GB array.)
When the server reboots it is highly unstable and a lot of services won't start (e.g. ISA, Exchange, SQL).
I have full C partition (OS location), system state, Exchange MDBDATA and SQL backup conducted by NTBackup for consecutive days before the blue screen.
The C partition is on a RAID5 volume (CVOL) which has a SMART error and bad stripes. This is what I have done.
1) I have created a new RAID1 volume (DVOL) of the same size as CVOL on new disks attaached to the same RAID controller. It's a RAID1 because I only had 2 free ports on the controller.
2) I booted the server in safe mode and, using server management console, created a new D partition on the DVOL volume. I formatted it.
3) I then rebooted the server in DSRM safe mode and used NTBackup to recover the C partition to the new D partition, and system state from the NTBackup from two days before the blue screen. I chose the option to replace all of the files on the destination (although it was empty of course) and maintain load points (not sure about this).
4) I rebooted the server, entered the RAID array control utility (Ctrl-A as it's an Adaptec array) and promoted DVOL ahead of CVOL in the boot order, restarting the server when the change was accepted.
Problem. The server doesn't boot into SBS 2003 R2. I just get a blinking cursor. F8 to select the boot options such as safe mode doesn't work. I have to power down and reverse point (4) to get an OS load.
Question 1. What am I doing wrong?
Question 2. How do I create a complete restore of the OS partition and system state but to a new Volume on the same hardware, so that I can avoid the SMART error and bad stripes that plague the original Volume?
Question 3. Is there anything specific I need to do so that I can eventually recover Exchange and all my SQL data bases?
Queston 4, alternatively, can I restore/repair to the original C partition on CVOL if I run chkdsk /r by booting the server from disk 1 of the SBS install CD-ROM and chosing to repair an existing installation? (For information, chkdsk /r has been running for 12 hours and has reached 54% on step 4 of 5. There are approximately 100,000 files on the C partition and they ocupy 140GB. CVOL is a 245GB array.)
ASKER
Thanks, David, I'll try this now. I'll let you know how it goes.
chkdsk /r completed in 18.5 hours saying it has found and fixed problems. I'm going to see whether the system now boots properly from the original C partition and, if not, go down this route you suggest.
Finally, the problem seems to have started when a power outage caused an uncontrolled shutdown (UPS didn't have enough juice to allow a clean shutdown) which lead to the bad stripe emerging on the array. Buying a more juicy UPS.
chkdsk /r completed in 18.5 hours saying it has found and fixed problems. I'm going to see whether the system now boots properly from the original C partition and, if not, go down this route you suggest.
Finally, the problem seems to have started when a power outage caused an uncontrolled shutdown (UPS didn't have enough juice to allow a clean shutdown) which lead to the bad stripe emerging on the array. Buying a more juicy UPS.
ASKER
Hello David,
Sorry for the delay in getting back to you. I have discovered that the bad stripes on my CVOL array, containing my C partition, have made a mess of the current SBS 2003 installation.
What I was planning to do was the following.
a) Remove the disks that make up the entire CVOL array and replace them with new disks.
b) Create a new CVOL array of the same size as the old one and reinstall SBS 2003 R2 from the CDs.
c) Reinstall the relevant service packs.
At (b) I was hoping I could avoid going through the whole installation process (NICs, IP addresses, Email settings etc etc) as I want to then recover all of the original C partition content from NTBackup.
Is that what you are saying I should do in your post? Do I abort the installation after the initial setup is complete and before all the questions that the wizard asks after the first reboot?
Sorry for the delay in getting back to you. I have discovered that the bad stripes on my CVOL array, containing my C partition, have made a mess of the current SBS 2003 installation.
What I was planning to do was the following.
a) Remove the disks that make up the entire CVOL array and replace them with new disks.
b) Create a new CVOL array of the same size as the old one and reinstall SBS 2003 R2 from the CDs.
c) Reinstall the relevant service packs.
At (b) I was hoping I could avoid going through the whole installation process (NICs, IP addresses, Email settings etc etc) as I want to then recover all of the original C partition content from NTBackup.
Is that what you are saying I should do in your post? Do I abort the installation after the initial setup is complete and before all the questions that the wizard asks after the first reboot?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
This does work. If there are problems in creating a new C partition, regedit.exe and a reboot fixes that, but everything else works fine. Thanks.
There is a bootstrap file installed on the first logical volume (someone with more knowledge will correct my terminology), without it, the system won't read BOOT.INI and boot your system. That file won't be on your new C Vol (but it is still on your old C Vol). I'm sure there are fancy ways to restore it, and there are 3rd party tools that can restore it, but with what you have, the way I have got around this is to start a fresh SBS install but only install Windows (don't go on with the Active Directory & SBS step), BUT, instead of accepting the default directory of C:\Windows, I stipulate to install it to C:\EMGWIN (emergency windows). If nothing else this always gives you a 2nd basic Windows O/S you can boot in emergencies. This process takes all of 25 minutes and does restore your bootstrap file on your new C Vol. Once this basic install comes up, you can edit BOOT.INI to put your real Windows as first boot option. This gets your Windows up again. So you aren't doing anything wrong, you just don't have the necessary boot file on the (now) first volume.
If you get this up and have done/do a full restore, then Exchange, SQL, etc should all come back too. You may need to run some EXEUTIL options on your Exchange database, or you may be lucky and it works first try. There will be postings about what to do to recover exchange after a rebuild elsewhere on EE.
As to the CHKDSK running on your old C Vol, it sounds like that volume is cactus in its present state. You will need to work out whether the issue (and root cause) is hardware, or just a corruption. If it's hardware, obviously that needs to be resolved. If it was just a random corruption (power fail, etc), then you should be able to reformat and restore to that volume again. Once you get it back up and running reliably again, invest in ShadowProtect for SBS. It will make such recoveries so much quicker/easier for you.
Hope the above is of some assistance,
David