Recovering an Exchange 2007 Clustered Mailbox Server (CMS) using the RecoverCMS Switch Part 2

If you missed the first part in this article series please read Recovering an Exchange 2007 Clustered Mailbox Server (CMS) using the RecoverCMS Switch (Part 1)

 

In the previous article in this two part article series on how you recover a lost CMS in Exchange Server 2007, we went through how you install the first passive mailbox server, recover the CMS itself and how you restore the Mailbox databases from backup. In this article we’ll continue where we left off. We will install the Passive Mailbox Server role on additional nodes in the Windows cluster, update storage group copies, resume replication, testing cluster fail-over and finally explain how you clean up the obsolete computer account objects and A-records in Active Directory and DNS respectively.

 

Installing the Passive Mailbox Server Role on additional Nodes

 

We now have a fully working Exchange 2007 cluster recovered, but since there’s only one node so far, it doesn’t provide much redundancy yet! So let’s get going and have the any additional cluster nodes added to the Clustered Mailbox Server. Adding additional nodes to the CMS is a straightforward task as it’s simply a matter of running the Exchange Setup wizard and then install the Passive Clustered Mailbox Role on each node. The procedure is identical to the one provided in the “Installing the Passive Mailbox Server Role one of the New Nodes” section earlier on in this article, so I won’t go through the steps again.

 


Figure 2.1:
Installing the Passive Mailbox Server Role on additional cluster nodes

 

Updating the Storage Group Copies on the Passive Node

 

Alright there are only a couple of steps more to go through before we can call the disaster recovery a success. Currently replication to the Storage Group copies on the passive node is in a suspended state. Figure 2.2 is a screenshot of the Mailbox database and storage group path on the passive node in my lab environment (yes since this is a test environment I can keep all data on the same disks, but you should of course never to do in a production environment) and as you can see it doesn´t contain any EDB databases or log files as, like I mentioned, replication is suspended.

 


Figure 2.2: Storage Group Folders still empty as they are still suspended

 

To resume replication, we should update the storage group copies on the passive node. To do so, open the EMC on the passive node, then navigate down to Server Configuration > Mailbox and select the Storage groups one at a time and click Update Storage Group Copy in the Action pane. In the Update Storage Group Copy wizard (see Figure 2.3) tick Delete any existing log files in target path, then click Next and then Update.

 


Figure 2.3: Updating the Storage group Copies on the passive node

 

When the respective Storage group copy has been updated successfully, it will be resumed automatically.

 


Figure 2.4: All Storage Group Copies have been updated on the Passive node

 

When the wizard has completed perform the same step for the rest of the suspended Storage groups.

 

Note:
you can of course also update all Storage group copies at the same time by use the Update-StorageGroupCopy CMDlet in the Exchange Management Shell (EMS). Like is the case with most wizards in the EMC, the Completion page in the Update Storage Group Copy wizard reveals the Cmdlet and the associated parameters you can use to do this.
As you can see in Figure 2.5, the database and Storage group paths now looks much better, that is the EDB databases and log files have been replicated, so we again have a redundant Clustered Mailbox Server solution up and running.

 


Figure 2.5:
Log and EDB files now replicated to the respective Storage Group folders on the passive node

 

We can now move on the final task, which is to test whether a fail-over to the passive node works as expected.

 

Testing Fail-Over to the Passive Node

 

With the additional node added to the CMS and storage group copy replication resumed, let’s see if we can move the CMS to the passive node using the Manage Clustered Mailbox server wizard. To move the CMS to the passive node, open the Exchange Management Console then drill down to Server Configuration > Mailbox and click Manage Clustered Mailbox Server in the Action pane (Figure 2.6).

 


Figure 2.6: Clicking Manage Clustered Mailbox Server in the Action pane

 

On the Introduction page of the Manage Clustered Mailbox Server wizard, select the first option which is Move the clustered mailbox server to another node (Figure 2.7), then click Next.

 


Figure 2.7: Moving the Clustered Mailbox Server to the Passive node

 

Now click Browse and select the node to which the CMS should be moved, then enter a move comment (Figure 2.8) and click Next.

 


Figure 2.8: Specifying the target node and entering a move comment

 

When the CMS has been moved to the other node, click Finish to exit the Manage Clustered Mailbox Server wizard.

 

Note:
Moving a CMS to the passive node using the Exchange Management Console (EMC) is only possible if you have installed Exchange Server 2007 Service Pack 1 on the nodes. If you still use the Exchange 2007 RTM version, you must instead use the Exchange Management Shell more specifically the Move-ClusteredMailboxServer CMDlet. The full command looks like the following: Move-ClusteredMailboxServer <CMS name> -TargetMachine: <passive node name> -MoveComment “note on why the CMS is moved!”

 

Let’s open the property page for the CMS in the EMC and click the Clustered Mailbox Server tab (Figure 2.9), and verify the once passive node (in this case CCR4) now is the Active node.

 


Figure 2.9:
Second node (CCR4) node the Active node

 

We can of course also verify where the CMS is located using the Cluster Administrator, but I thought I wanted to show you one of the new cool improvements in Exchange 2007 Service Pack 1 J

 

Cleaning Up Active Directory and DNS

 

We have now completed the disaster recovery steps and everything looks good. The final thing to do is to clean up the Active directory as well as DNS, by removing the now obsolete computer account objects, Windows cluster node and DNS A records.

 

To remove the old cluster node computer account objects from Active Directory, open the Active Directory Users and Computer MMS, and then select the Computers organization unit (OU). Find the respective computer account objects, and delete them by right-clicking each and select Delete in the context menu (Figure 2.10).

 


Figure 2.10: Deleting the Computer Accounts for the old Cluster Nodes

 

To dele the obsolete DNS records, open the DNS Management console and then select the respective forward lookup zone. Find the records and delete them by right-clicking on each and select Delete in the context menu (Figure 2.11).

 


Figure 2.11:
Deleting the DNS A records for the old Cluster Nodes and the Windows Cluster
That was it! Doesn’t it feel great to actually have been through the steps necessary to recover a CMS?

 

Conclusion

 

Since messaging systems are considered mission critical in most organizations today, many organizations deploy redundant messaging solutions in order to eliminate single point of failures. But even a redundant system such as an Exchange 2007 Single Copy Cluster (SCC) or Cluster Continuous Replication (CCR) based Clustered Mailbox Server (CMS) can be lost (for example by means of earthquakes and other natural disasters). If you end up in a disaster recovery situation its vital you know how to recovery from the situation and this is only possible if you have performed extensive testing prior to the disaster.
In this article we covered how you can recover a CMS. And as you have seen throughout the article, the process for recovering a clustered Exchange Server 2007 Server is much less complex than was the case with Exchange Server clusters in previous version of Exchange Server. But this doesn’t mean you shouldn’t test how to recover a CMS as part of your disaster recovery plan.

 

If you missed the first part in this article series please read Recovering an Exchange 2007 Clustered Mailbox Server (CMS) using the RecoverCMS Switch (Part 1)

About The Author

Leave a Comment

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Scroll to Top