These instructions are for installing the Vault and/or an Agent on each of the cluster nodes. You will have an identical installation on each node. You can install the Vault and/or Agents on Microsoft Windows Cluster platforms. You can cluster just the Vault, just Agents, just some Agents, or everything.
Scroll down past the procedure for information about Jobs on colocated Agents and remote Agents that are not part of the cluster.
Before you begin, review the following information:
Ensure that a shared disk is set up between cluster nodes, outside of the cluster computers, such as a NAS. This disk will hold the Vault data that is shared between clusters.
Do not create any Jobs until after configuring a Vault on each node of the cluster.
After you have set up your cluster software and identified a shared disk that each cluster node can share, you are ready to install the WAFS software. For assistance installing and configuring a failover cluster on Windows 2008, refer to the Microsoft Technet article Failover Cluster Step-by-Step Guide: Configuring a Two-Node File Server Failover Cluster.
To install or upgrade the Vault and/or Agent on each of the cluster nodes
On the active node, run the WAFS installer, following the procedure in Installing WAFS or Upgrading the Software to install both the Vault and an Agent or just the Vault.
Be sure to specify the shared storage location that you set up for the cluster nodes on the Configure WAFS Vault page, in the Working Directory and Control Vault area. You must point each clustered Vault to the same shared storage location.
The illustration below is an example of setting up the shared storage (disk) location.
Once installation is complete, prior to a system reboot, configure the WAFS Vault and/or Agent services to start Manually, as this will ensure proper WAFS behavior in a clustered environment. (Your cluster software might set this up for you, but verify in the Services snap-in that the service is set to start Manually, not Automatically, before you reboot.)
Stop the cluster service on this node, then make the second node active.
Run the WAFS installer on the second (now active) node, installing the same components (Vault and/or Agent), using the same shared storage location, and setting the services to start Manually. The nodes must be identical for smooth failover.
Set up a "Generic Service" in the cluster for the Vault service (you can do this on either node):
In the configured clustered environment, click Manage a Cluster.
Select the cluster that will be used for WAFS.
Under Services and applications, click Configure a Service or Application.
Click Generic Service, then click Next.
In the list of available services, click WAFS Vault, then click Next.
Provide a descriptive name for the service such as WAFSVaultServ. (The length of the name is limited; you will receive an error if the name is too long.)
Specify the IP address that will be shared between cluster nodes, then click Next.
On the Select Storage page, assign the shared disk, then click Next.
On the Replicate Registry Settings page, click Add, then specify one of the following registry locations, depending on whether the system has a 32-bit or 64-bit OS. (The dialog box provides the first part of the key.) IMPORTANT: This registry location must be specified correctly. You can avoid errors by copying and pasting the key from the Registry Editor or from the text below.
Complete the wizard.
Verify the same service appears in both nodes of the cluster.
Verify the active node automatically started the WAFS Vault service.
If you have also clustered the Agent, repeat the Generic Service procedure for the WAF Agent service.
For Jobs on colocated Agents:
When creating a WAFS Job on an Agent that is colocated with the Vault, use "localhost" for the Vault name so that WAFS will connect to the local computer regardless of the node. You can instead use the shared cluster IP address, however, performance is improved when using localhost rather than an IP address.
After creating a Job, verify that the AVMF folder is automatically created on the cluster's shared storage disk.
For remote Agents that are not part of the cluster:
When a Cluster active node fails, it takes ~60 seconds to time out and quit its operation.
You can configure the Agent to wait for a clustered disk by setting the following registry location
to 1 (create the key if does not exist). (Editing the registry is for advanced users.)