I have an immediate need to replace a pair of 2003 Enterprise servers acting as a clustered file server with a standalone 2008 file server.
We've been migrating user data off of this cluster, but the underlying disk is at risk.
I've done this before with file servers (and a SQL server) by doing the following:
1. bring up the new server and storage and join to the domain
2. restore seed data from last full backup.
3. sync data from production system to new system
4. Export share maps from production system registry
5. Import share maps on "new" file server
6. disconnect production system
7. assign IP from the (now old) production server to the new production server (to accomodate any raw ip mapping)
8. drop the NEW system from the domain, change the name to match the OLD system, and re-add to the domain
The above steps have worked flawlessly in the past with single servers and even a SQL server. What I'm uneasy about is the way that server 2003 clusters interact with DNS and AD. Specifically the way they handle the cluster object vs. unique cluster member objects, and handling a roll-back scenario.
Before we begin, I'll be shutting down the passive node. Is there anything else I need to know about how the cluster object is treated in AD? With the active node powered on but disconnected from the network, will I still be able to join the new standalone server to the domain using the same name?
If not, are there specific objects I need to remove from AD prior to joining the new server? I'd like to be able to shut down the new system, restore AD objects, and reconnect the legacy system in a 'worst case scenario' situation.