In this document, a disaster is defined as having to restore Exchange Server and/or Windows 2000 Server. With Exchange 2000 the introduction of multiple storage groups and databases adds more complexity to restoring. But in addition to using Windows 2000 Backup to back up and restore, you can reinstall Exchange 2000 using the /DisasterRecovery option. This option allows you to run Setup in Disaster Recovery mode to rebuild a server previously lost in the Exchange topology.
Besides Disaster Recovery setup, there are other procedures needed for recovering servers. In Exchange 2000, servers take on different roles such as Key Management Server and SRS. In addition to these new server roles, the platform Exchange is running on also adds more steps. Recovering an Exchange 2000 cluster server requires more steps than recovering a single Exchange 2000 member server.
This section describes recovery requirements and steps in the following scenarios:
There are five common requirements for recovering all Exchange 2000 servers:
Important These steps assume that your Exchange 2000 server is a member server in a domain and not the server that runs Active Directory.
If the Exchange 2000 server is also an Active Directory server be sure to include backups of Active Directory in the system state backup. For information about the requirements for restoring Active Directory on a domain controller, see the Windows 2000 Server Operations Guide.
If Active Directory runs on a separate domain controller in the domain and it is intact, proceed with the steps below. If Active Directory is running on the same computer as the Exchange 2000 member server, you must first restore Active Directory before restoring Exchange 2000. For more information about restoring Active Directory, see the Windows 2000 Server Operations Guide.
In addition to reinstalling Windows 2000 on the computer and restoring file system backups, it is recommended that you restore the Windows 2000 system state. When you complete a backup of the Windows 2000 system state, which includes the Windows registry and IIS metabase, the computer returns to the state it was in before the backup.
You might need to reinstall Windows 2000.
To reinstall Windows 2000
Important After reinstalling the correct version of Windows 2000, reinstall any service packs or hot fixes that were previously installed.
You should restore full backups of the system drive or any other logical drives where critical application data was installed. Use Windows 2000 Backup to restore file system backups to the computer on which you restored Windows 2000.
When you restore the Windows 2000 system state, the restored computer returns to its original domain where its computer account matches the System ID (SID) in Active Directory. Use Windows 2000 Backup to restore the Windows 2000 system state. After the restore, Windows 2000 Backup prompts you for the computer you want to restart.
Important After restoring the system state and restarting the server, you might see an error message indicating that one or more services cannot be started. This includes services such as SMTP that were running prior to restoring system state, and services that have not been installed yet. These services require that you install Exchange 2000. If the full file system restore does not include the Exchange 2000 installation directory or other critical program data, this error message occurs. These services start after Exchange 2000 installs in disaster recovery mode.
Following the restore of your system state, the event log might show that some Exchange 2000 services have failed. If these services are not installed yet, when you restore the system state, Windows 2000 accepts that these services are installed on your server. These services start after Exchange 2000 installs in disaster recovery mode.
When you run Setup.exe with the /DisasterRecovery option, Exchange 2000 restores executable files and system settings without disturbing the existing Active Directory information for the system. Setup in disaster recovery mode installs Exchange 2000 without resetting the server’s configuration to defaults, but instead, leaves the server in its last configuration.
To run Exchange Setup in disaster recovery mode
Important You must install Exchange 2000 to the same drive and directory on which it was installed on the original server.
To recover databases in Web Storage System, you must verify that all services on which Exchange 2000 depends are running. To restore a database, you must also dismount the database. However, because Exchange supports multiple storage groups, you only need to dismount the specific database you want to restore. This allows users access to any other databases in Web Storage System.
To recover databases in Web Storage System
Recovering an Exchange 2000 member server also running Site Replication Service (SRS) requires the same steps involved in recovering a single member server. However, there are additional steps to recover the SRS database after you run Disaster Recovery setup.
In addition to the requirements described in “Requirements for Recovering Exchange 2000” earlier in this chapter, back up the Exchange 2000 SRS database. Follow the same steps listed in “Recovering an Exchange 2000 Member Server” earlier in this chapter. Reinstall Windows 2000, restore the system drive backup, restore the Windows 2000 system state backup, run Exchange 2000 Setup in disaster recovery mode, and restore Exchange Web Storage System databases. However, prior to restarting the computer as described in the last steps listed in the “Recovering Databases” section of “Recovering an Exchange 2000 Member Server” earlier in this chapter, you must perform the steps listed below to recover the SRS database from backup. Recover the SRS database using Windows 2000 Backup, and then restart the server.
To enable and start SRS after disaster recovery
To reset the password for the Exchange 5.5 service account
Using Windows 2000 Backup, follow the same steps listed in “Recovering an Exchange 2000 Member Server” earlier in this chapter. However, instead of selecting Exchange Information Stores with Ntbackup, select the SRS database to be restored.
When you restore the SRS database, you have completed the recovery of the Exchange 2000 member server running SRS. Before you restart the system, verify that you have performed all the steps described in “Recovering an Exchange 2000 Member Server” earlier in this chapter.
Recovering an Exchange 2000 member server also running Key Management Service requires the same steps involved in recovering a single member server. However, there are additional steps to recover the Certificate Authority (if Certificate Authority was running on the same server as Key Management Service) and Key Management Service database after you run Setup in disaster recovery mode.
In addition to the requirements described in “Requirements for Recovering Exchange 2000” earlier in this chapter, make sure you have the following backups:
Follow the same steps listed in “Recovering an Exchange 2000 Member Server” earlier in this chapter. Reinstall Windows 2000, restore the system drive backup, restore the Windows 2000 system state backup, run Exchange 2000 Setup in disaster recovery mode, and then restore Exchange databases.
If the recovered server was also the Certificate Authority, in addition to being the Key Management Service; then by restoring the system state of the original server, Certificate Authority is also restored. Certificate Authority does not have to be checked as a component to install if you are restoring the system state.
Prior to restarting the computer as described in the last steps of “Recovering Databases” in the “Recovering an Exchange 2000 Member Server” section earlier in this chapter, perform these additional steps in recovering the Key Management Service database from backup. Then restore the Key Management Service database using Windows 2000 Backup and reboot the server.
To start Key Management Service following disaster recovery
Use Windows 2000 Backup to perform the steps as described in the “Recovering Databases” section in “Recovering an Exchange 2000 Member Server” earlier in this chapter. However, instead of selecting Exchange Information Stores with NTBackup, select the Key Management Service database you want to restore.
The Exchange 2000 member server running Key Management Service recovery is complete after you restore the Key Management Service database. Before you restart the system, verify that you completed the steps described in “Recovering an Exchange 2000 Member Server” earlier in this chapter.
Clustering provides a mechanism for moving resources between cluster nodes when a disaster occurs. In the case where a single node fails, clustering moves Exchange 2000 resources to another node in the cluster so that services remain available to users. The node that failed can be removed from the cluster and then later replaced with another node joining the cluster. Exchange resources can then be moved back to the newly joined node so that load balancing is again achieved. This section lists the steps involved in removing a non-functioning node from a cluster, rebuilding, and then rejoining the node to that cluster.
In addition to disasters involving the loss of a single node in a cluster, there are cases where the cluster-shared disk is lost. This section describes how to recover when the cluster quorum is lost.
Clustering provides recovery when a server node goes down in a cluster. When a single node in a cluster fails, Exchange resources running on the node are moved to an available node in the cluster. Exchange databases remain intact on shared storage and can be accessed by the Exchange virtual server from another node in the cluster. This provides reliability when disaster occurs on a single node in the cluster. Once resources are moved to an available node in the cluster, follow these procedures for removing the non-functioning node and replacing it with a new node.
When one cluster node suffers a disaster and needs to be replaced by a new node, you must evict the lost node.
To evict the lost server node from the cluster
The lost node does not have to be rebuilt like the original lost node. An entirely new node can be built (new computer name, new IP) and then joined to the cluster. Perform the following steps to build a new server node.
To build a new server node for the cluster
To rejoin the server node to the cluster, set up Cluster service on the newly built server. When asked to join a cluster, specify the cluster you want to join.
You might need to install Exchange on the server node and move resources back to the node.
To install Exchange on the server node and move resources back to the node
In order to recover from a cluster quorum failure, you must perform a cluster quorum backup. In addition to the requirements described in “Requirements for Recovering Exchange 2000” earlier in this chapter, you must also have a system state backup containing the cluster quorum.
Typically you need to recover a single lost node in a cluster. However, you might have a case where the cluster quorum is lost on the shared disk along with the Exchange databases. If this occurs, you must restore the cluster quorum from backup.
Before you can restart Cluster service on any nodes in the cluster, you must restore the quorum.
To restore the quorum from backup
After you restore the quorum and restart the nodes in the cluster, verify whether the shared disk resource can be accessed after the Cluster service has started. If the shared disk where Exchange databases reside can be accessed and has survived the disaster, check to see if the .edb, .stm, and .log files still exist for the Exchange virtual server storage groups. If they are intact, start your Exchange resources. If the shared drive is lost, restore your Exchange databases from backup.
To restore Exchange 2000 databases from backup
Note On a cluster server, you must verify that the shares where Exchange databases reside are available to and accessible by the cluster node that owns the disk resource.
Source: Exchange 2000 Resource Kit