Enterprise Vault migration/Exchange Upgrade/EV upgrade

Hello Kristjan,

I'd go another way:

Upgrade vault to EV9
'upgrade' to Exchange 2010 SP1 (!!!!) You need SP1 to have EV9 archive from it.
Re-install nodes with 2008r2

This will give some 'downtime' whilst upgrading the cluster EV runs on. But, if you use the same names, provided you have the aliasses setup in DNS, you could be up and running reasonably quick.

Another way, as you say you want to 'restore all data to exchange', if you have the diskcapacity to do so:
Disable archiving for all users
Restore all archived items back into exchange
uninstall/remove EV8
upgrade to Exchange 2010
New cluster, new EV
Archive

Be aware that in the last scenario, you need to think (amongst others) what to do with shortcuts in mailboxes (if you use those) and in the PF's.

There might be other options, but perhaps you should consult with symantec to see what they think.

I agree with Gertjan,

You should upgrade to EV9 first. Next upgrade your Exchange environment to 2010 SP1

Next do a Data only DR on the EV servers by backing up all the data and indexes, Install 2008 R2 64Bit (must be 64Bit as 32Bit is not supported with EV), restore data and indexes to new cluster(if restore is needed) install EV in a new cluster.

The Data only DR process is detailed in the Admin guide and it works great

Thank you for your reply's

Gertjan, if i choose the "export" option and choose "restore items to the original mailbox", doesnt it replace shortcuts with original items?
I cant do it with public folders tho... so this option goes to the bin.
 
If the steps are:
1)Upgrade to EV9
2)Upgrade to exchange 2010 sp1
3)Reinstall 2008R2 on EV nodes (witch breaks the cluster) and EV
4)create new cluster with same properties and install EV
5) restore

about 5) - when i have all the indexes, vault stores, Shopping on storage disks and sql database on different server. Is there a possibility to "reconnect" when i'm reinstalling EV?

Thank you,
Kristjan

Hi Kristjan,

On your question: yes.

On 5 - yes. Check the Data only DR (disaster recovery) as indicated by Liam.

Good luck. If you go this way, I can add 1 advise: backup, backup, backup.
Backup sql db's, data, indexes etc. Make sure they are good.

I learned the hard way....

Hi again. This are the steps for rebuilding a cluster. I assum restoring the sql is not necessary, as you have those available.

Scenario 2: All the cluster nodes need rebuilding
If all the cluster nodes need rebuilding, but you have backup copies of the SQL Server databases and the Indexing and saveset data for Enterprise Vault, you can follow the steps below to rectify the situation.
To repair all the nodes in the cluster
1 Rebuild all the computers and restore the SQL databases as follows:
■ Restore the Enterprise Vault Directory database.
See “Procedure 2: Restoring the Enterprise Vault Directory database” on page 330.
■ Restore the vault store databases.
See “Procedure 3: Restoring the vault store databases” on page 331.
■ Restore the Monitoring database.
See “Procedure 4: Restoring the Monitoring database” on page 332.
■ Restore the FSA Reporting database.
See “Procedure 5: Restoring the FSA Reporting database” on page 334.
2 Recreate the cluster using Microsoft Cluster Administrator. Use the same number of disks, and mount the shared disks with the same drive letters as before.
3 Restore the Indexing and Enterprise Vault store data to disks using the same drive letters as before.
See “Procedure 7: Copy or move the Enterprise Vault data files” on page 337.
4 Recreate the resource groups, including the prerequisite resources, using the original virtual server names.
5 Install Enterprise Vault on all the nodes in the cluster.
6 Run the Enterprise Vault Configuration wizard on each primary node. Choose to configure a new Enterprise Vault server with cluster support. The wizard detects the existing virtual server name in the Enterprise Vault Directory
database’s ComputerEntry table, and performs a repair. When the repair is complete, a wizard page is displayed, with which you can recreate the Enterprise Vault service resources and Server Instance resource.
Do not bring the resources online when given the option to do so.
7 Run the Enterprise Vault Configuration wizard on each failover node. Choose to add the node as a failover node for an existing clustered server.
8 Clear the Directory database entries.
See “Procedure 8: Clearing the directory database entries” on page 338.
9 Check the Web Access application URL.
See “Procedure 11: Checking theWebAccess application settings” on page 342.
10 Bring the cluster resources online and test that failovers work as planned.

Hi Kristjan,

Backup just in case absolutely.
2,3,4 yes

I am not sure on the order of 5,6,7. Recalling repairing a faulty upgrade, I believe that step 7 (clearing directory database entries) should be done BEFORE step 5 and 6.

What I would do is follow the steps as outlined in the document. When the configuration wizard runs without problems, recreating the services and entries etc, and you can start the console, and client-functionality is available and report runs of archiving task and provisioning are correct, I don't see the need to empty the directory database entries. Perhaps you can verify with support the exact order of the steps.

Thank you for your help:)

Now i have actionplan in place, i'll try to confirm it with Symantec also:)

Kristjan