windows 2003 cluster secondary node disaster recovery

[Kurt Beyers.]

Hello,

I've got a Windows2003 F&P cluster (no DC's) and we are testing the disaster recovery of a secondary node of the cluster. The primary node still has the quorum disk and all of the shared disks. The backup is taken with DP5.1

I would use the following procedure for the recovery/restore:

1. install a fresh OS
2. install the DP client software
3. perform a full restore of the local disks (overwrite and move busy files)
4. perform a configuration restore, these are all the Windows System state object.
5. Reboot the server

Do you have to restore the Cluster Database as well in the configuration restore? Or is this (automatically) recovered from the primary node when the secondary node joins the cluster again?

best regards,
Kurt

[Rune J. Winje · Accepted solution]

Sounds OK to me.

No need to specifically restore the cluster db - the restored node will sync up.

Also evicting the old node from the cluster and joining with a new node in the cluster is another way to do it.

Further there's a brilliant tool in the Win2K3 reskit called clusterrecovery.exe which addresses the other #1 topic - when your SAN-disks are shot/corrupt and need replacement and you put in new disks with restored data. (Can also be used for planned disk-replacement).
http://www.microsoft.com/downloads/details.aspx?familyid=9d467a69-57ff-4ae7-96ee-b18c4790cffd&displaylang=en

DO NOT use the Clustool.exe included with Win2K reskit, as that stopped working with SP3 for Win2K. Clusterrecovery.exe however is backported for use with Win2K clusters, but must probably be run form Win2K3 or WinXP machines.


Cheers,
Rune