Exchange 2003 Disaster Recovery - Restoring to a Single Server from a failed Cluster
Ok, so many of us out there are faced with situations which require us to restore data backed up from an Exchange Cluster onto a single server.
This can happen in multiple scenario's, maybe your SAN has become unavailable for some reason, maybe you need to test your DR Procedure in a lab environment to make sure your prepared in the event of a Disaster & that your backups are indeed reliable.
There are many reasons out there but not so many reasonable howto's.
Im going to try give you a general idea here on the procedure - dont expect a 100% accurate step by step, its more of a general guideline which if you use a bit of common sense, intelligence and well, some more research you should be able to perform the D/R with no problems at all.
Onto the Procedure:
Delete the Cluster Virtual Server object from A/D using ADSIEDIT
Delete the Virtual Server Computer object from Active Directory Users & Computers
Install the new server with the same server name, O/S Level & Patches then install a fresh copy of Exchange as part of the origional organization
You will need to create the exact same drive & folder structure as what you had on the cluster, e.g. x:\exchsrvr\mdbdata\priv1.edb etc
Create Identically named Storage Groups & Information Stores which are located in the same path as on the cluster then dismount the stores & mark them to never be mounted @ startup & Database may be over written by a Restore
Add the following Registry DWORD Recovery SG Override = 1 to this registry key:
HKLM\SYSTEM\CurrentControlSet\Services\MSExchangeIS\ParametersSystem
(Remember to set this back to 0 after the restore)
Install your backup software, catalogue the last successful backup media & restore your information stores.
* It is normally a best practice when restoring multiple information stores to run each store as an individual job - this allows them to be mounted one at a time as and when the restore completes.
Now reconnect all the mailboxes to their respective user accounts using the Mailbox Recovery Centre in Exchange System Manager & your system is back up & running.
This can happen in multiple scenario's, maybe your SAN has become unavailable for some reason, maybe you need to test your DR Procedure in a lab environment to make sure your prepared in the event of a Disaster & that your backups are indeed reliable.
There are many reasons out there but not so many reasonable howto's.
Im going to try give you a general idea here on the procedure - dont expect a 100% accurate step by step, its more of a general guideline which if you use a bit of common sense, intelligence and well, some more research you should be able to perform the D/R with no problems at all.
Onto the Procedure:
Delete the Cluster Virtual Server object from A/D using ADSIEDIT
Delete the Virtual Server Computer object from Active Directory Users & Computers
Install the new server with the same server name, O/S Level & Patches then install a fresh copy of Exchange as part of the origional organization
You will need to create the exact same drive & folder structure as what you had on the cluster, e.g. x:\exchsrvr\mdbdata\priv1.edb etc
Create Identically named Storage Groups & Information Stores which are located in the same path as on the cluster then dismount the stores & mark them to never be mounted @ startup & Database may be over written by a Restore
Add the following Registry DWORD Recovery SG Override = 1 to this registry key:
HKLM\SYSTEM\CurrentControlSet\Services\MSExchangeIS\ParametersSystem
(Remember to set this back to 0 after the restore)
Install your backup software, catalogue the last successful backup media & restore your information stores.
* It is normally a best practice when restoring multiple information stores to run each store as an individual job - this allows them to be mounted one at a time as and when the restore completes.
Now reconnect all the mailboxes to their respective user accounts using the Mailbox Recovery Centre in Exchange System Manager & your system is back up & running.
Comments