Cluster resources can be configured to restart automatically

This article describes how to configure cluster resources to restart automatically in Windows 2000 Server, in Windows Server 2003, and in Windows Server 2008.

When a cluster resource encounters a problem and then fails in a health check, the cluster resource enters a failed state. In this failed state, the resource tries to restart automatically until it reaches the restart threshold that is defined by the properties of the resource.

In Windows 2000 Server and in Windows Server 2003, the resource remains in a failed state when the threshold is exceeded. In this case, the resource waits for manual intervention.

Note The default restart threshold in Windows 2000 Server and Windows Server 2003 for all resources is three failures over a period of 900 seconds. In Windows Server 2008, the default restart threshold is one failure over a period of 900 seconds.

For more information about the changes for failover clustering in Windows Server 2008, click the following article numbers to view the articles in the Microsoft Knowledge Base:

The RetryPeriodOnFailure property

A cluster resource may have problems because of transient conditions in the environment. In these situations, you may want to restart later. In Windows 2000 Server and in later versions of Windows Server, each cluster resource has a property that defines how long to wait after the resource has exceeded its restart threshold before Windows Server tries to bring the resource online again. This property is called RetryPeriodOnFailure.

In Windows 2000 and in Windows Server 2003, the default value of this property is "disabled" (0xFFFFFFFF). In Windows Server 2008, the default value of this property is 60 minutes.

When the RetryPeriodOnFailure property is set to 0xFFFFFFFF, the Cluster service never tries to automatically restart a resource when the resource is in a failed state.

Note In Windows 2000 and in Windows Server 2003, you can modify the RetryPeriodOnFailure property only by using the Cluster.exe command-line tool.

In Windows Server 2008 failover clustering, you can modify the RetryPeriodOnFailure property in the Properties dialog box of a resource. To do this, follow these steps:

1. Start Cluster Administrator, right-click the resource that you want to modify, and then click Properties.
2. In the Properties dialog box, click the Policies tab.
3. Enter the desired time in the box that corresponds to the If all the restart attempts fail, begin restarting again after the specified period (HH:MM) option, and then click OK.

How to use the Cluster.exe command-line tool to modify the RetryPeriodOnFailure property

To view the current property value, type the following command at a command prompt, and then press ENTER:Note In this command, clustername is a placeholder for the name of the cluster to which you want to connect, and resourcename is a placeholder for the name of the resource that you want to modify.

To modify the RetryPeriodOnFailure property, type the following command at a command prompt, and then press ENTER:Note In this command, xxx is a placeholder for the time in milliseconds.

The value of the RetryPeriodOnFailure property must be greater than the value of resource's RestartPeriod property. By default, this value is 900,000 milliseconds or 15 minutes. (You can modify this value.)

We recommend that you use intervals that are measured in minutes (multiples of 0x0000EA60). When you do this, you make sure that the Cluster service does not restart a resource more than the number of times that is specified in the "RestartThreshold" value during the restart period.

For more information, click the following article number to view the article in the Microsoft Knowledge Base: