Generating Notifications for an MSCS Resource Problem (260527)

MORE INFORMATION

First, create the batch file to be used to perform the notification. In this example, a net send command is used. Or, you can use Logevent.exe from the Microsoft Windows NT 4.0 Resource Kit or the Microsoft Windows 2000 Resource Kit to log events. Create a file named Notify.bat with the following text: NOTE: If you want multiple users to get an notification, duplicate the START /B line, and then list another contact person.

Next, create a Generic Application resource in the appropriate group:

1. Right-click the approrpriate group, point to New, and then click Resource.
2. Fill in the boxes, making sure to click Generic Application in the Resource Type box, and to click the correct group. Click Next.
3. Make sure both nodes are listed as possible owners, and then click Next.
4. Define the appropriate dependencies. At a minimum, define the physical disk on which the utility is located.
5. In the Command Line box, type the full path and file name of the batch file that you created. Type the full path in the Current Directory box (for example, T:\). Determine whether the Allow application to interact with desktop option is appropriate. For testing, you may want to enable this option initially. Click Next.
6. Skip the registry replication by clicking Finish.

Next, test the batch file:

1. Bring the resource you created online by right-clicking it in Cluster Administrator. The node that owns the group has a command prompt window if the Allow application to interact with desktop option is enabled, and a message is sent to the name specified in the net send command.
2. Right-click a monitored resource, and then click Initiate Failure. The node that owns the resource restarts the resource and resources that depend on it, resulting in another alert notification.
3. Right-click the group, and then click Move Group. The group should move and send a notification.

If you have a critical resource that you want to be independently monitored, create a dependency on that specific resource to ensure notification.

NOTE: You will be notified of every single problem, not just failovers. (By default, a resource must stop working three times before the group fails over to the other node.) You may see a large number of alerts with individual dependencies set.

For more detailed information about what causes a resource to stop working, see the event logs and the following Microsoft Knowledge Base article for information about cluster logging: