High Availability FAQsThis FAQ page is devoted to High Availability (and Clustering) related questions (primarily moved from Ask a Question page).
Q: Where can you find more information to troubleshoot an issue when failing over to a secondary cluster in Exchange 2007? It was working previously, and can still fail back to the primary server without issue.
Errors in Event Viewer:
Event Type: Warning - Event Source: MSExchange Cluster - Event Category: ExRes - Event ID: 1030
Description: Exchange Information Store Instance (EXCH-CL): Terminate process called for MSExchangeIS (ProcessId 10432).
Event Type: Error - Event Source: MSExchange Cluster - Event Category: ExRes - Event ID: 1003
Description: Failed to bring the resource Exchange Information Store Instance (EXCH-CL) online.
Event Type: Error - Event Source: MSExchangeIS - Event Category: General - Event ID: 5000
Description: Unable to initialize the Microsoft Exchange Information Store service. Failed to find the working directory parameter from the registry - Error 0x80004005.
A: This is a known issue with clusters in Exchange Server 2007. It has to do with the fact that ExRES is not honoring the default timeout values on the affected services. Instead it is appling its default value of 40 Seconds. Running the command "cluster res “<affected service name>” /prop PendingTimeout=150000" (e.g. "Cluster res "Exchange Information store Instance (MyServer)" /prop Pendingtimeout=150000") from a command prompt on the cluster should correct your problem. If it does not then you are hitting a larger issue and should contact Microsoft support to help you diagnose the issue.
Q: I currently am running a Exchange 2003 deployment in an Active/Passive configuration. I am planning to upgrade the cluster to Exhange 2007. I will evict the passive node from the Exchange 2003 cluster, install 2007 to it and migrate the mailboxes.
My question is, can the Exchange 2007 cluster name be renamed? I have read that the Exchange 2003 cluster name cannot be renamed. For example: Exchange 2003 cluster name (Mail), we are running Outlook 2003, so auto discovery is not avaiable as it is in Outlook 2007. We have 1000 Outlook users, so manually reconfiguring the Outlook profile is not an option. So, if I can rename the Exchange 2007 cluster name then the migration will be painless, install the 1st Mailbox server to a Active cluster node with a cluster name of (Mail1 for example). If the Exchange 2007 cluster name cannot be renamed, then I will need to bring up a Temporary Exchange 2007 server, move all 1000 mailboxes to it during off peak times, then un-install the Exchange 2003 cluster, and build a Exchange 2007 single copy cluster using the old Exchange 2003 cluster name (Mail in this example) and move the maiboxes back from the Temp Exchange 2007 install to the Exchange 2007 single copy cluster installation. Are there any suggestions?
A. When you move a mailbox between servers in an Exchange organization (even with 5.x servers) the client will get redirected to the correct Exchange server as long as the original server is still operational. So, you can evict the passive node, rebuild with Windows Server 2003 x64, build the single node Exchange Server 2007 cluster, test and move the mailboxes (you will also need a HT and CAS box to support mail access and delivery) and all legacy clients will redirect to the the new cluster automatically. Autodiscover is used to help Outlook 2007 initially get configured (and find all of the web services after configuration).
Q: What are the different backup strategies for Exchange 2007? We currently don't use MAPI mailbox level backups because of the slow speed. Instead, we use an agent based full store level backup for disaster recovery AND use exmerge to build mailbox level PST files for user level restores.
A: Exchange 2007 provides two technologies to do database backup/restore. One is the Legacy Streaming backups done at database level, this is the technology used by NTbackup. The other is VSS based backups done at storage group level, there is no in-the-box VSS based backup solution for Exchange (there will be one in LH server). Since MAPI is fully supported as a client protocol, it is possible to do mailbox level backups but there is no in-the-box solution to provide this functionality. Other then slowness of mailbox level backups, also keep in mind that mailbox level backups do not trigger log truncation and expiration of deleted items and deleted mailboxes. In replicated Exchange 2007 deployments (LCR and CCR), it is recommended to do weekly Full backups and daily Incremental backups from the Replica location, and continue doing daily Full backups for non-replicated Exchange databases. Mailbox level recovery can be achieved via database restores to RSG, and directly merging recovered content to any connected mailbox in the same Exchange organization.
Q: We are using CCR and the log shipping is very far behind. The queue does down at night but during the day it fills up again. Is there a way to get the logs caught up or to speed up CCR? We are trying to up our bandwidth but is there any thing else? Can we manually move logs? Thanks!
A: There's some good CCR tuning tips at: http://msexchangeteam.com/archive/2007/01/15/432207.aspx
| 
