Failover/Failback Policies on Microsoft Cluster Server

ID: Q197047

The information in this article applies to:

Microsoft Windows NT Server, Enterprise Edition version 4.0
Microsoft Cluster Server

SUMMARY

In Microsoft Cluster Server (MSCS) version 1.0, it is necessary to configure the "failover" and "failback" properties of the groups on the cluster.

MORE INFORMATION

Setting Failover Policy

To set the Failover policy for a group, use the Failover tab for the group. You can set the failover threshold and the failover period. The failover threshold is the number of times the group can fail over within the number of hours specified by the failover period. For example, if a group failover threshold is set to 5, and its failover period set to 3, Cluster Server will failover the group at most five times within a three hour period. The sixth time a resource in the group fails, Cluster Server will leave the resource in the offline state instead of failing over the group.

Setting Failback Policy

By default, groups are set not to failback. Unless you manually configure your group to failback after failover, it continues to run on the alternate node after the failed node comes back online.

When you configure a group to automatically failback to the preferred node, you specify whether you want the group to failback as soon as the preferred node is available or to failback only during specific hours that you define. This option is useful if you want the failback to occur after peak business hours, or if you want to make sure the preferred node is able to support the group when it does come back online.

The group must be configured to have a preferred owner to failback. You can specify a preferred owner on the General tab of the group Properties dialog box.

NOTE: The "Preferred Owner" of a group must be specified for failback to occur. The preferred owner is the node that one configures to (under normal operating circumstances) to host the group. Furthermore, on a resource level, the resource must be configured to have both nodes as "Possible Owners" for the resource to failover.

Testing Failover Polices

You can test the failover policies you establish for a single group and its resources by manually failing over those elements.

To test the failover policy for a group, type 0 (zero) in Threshold in the group Properties dialog box. Then, right-click a resource and click Initiate Failure. Cluster Server immediately fails over the group to the alternate node.

In a test environment, you can fail over all groups from one node to another by using Cluster Administrator to stop the Cluster Server, pressing the reset button on the computer, or turning off the power to one of the nodes.

NOTE: Removing the shared SCSI cable between either node and the shared disk array is not a valid failover test. The cluster service must have access to the shared array at all times for failover to succeed.

REFERENCES

For additional information on resources and groups, please see the following article in the Microsoft Knowledge Base:

Q169017 Information on Groups & Resources Using Cluster Server

Additional query words: mscs

Keywords :
Version : winnt:4.0
Platform : winnt
Issue type : kbinfo