To test node failure

• Test node failure using one of the following four methods:
• Stop the Cluster service on the node.

  This prevents clients from accessing cluster resources through that node. In this event, all resources owned by this node fail over to other nodes in the server cluster. For more information on stopping the Cluster service, see Related Topics.

• Turn off the power on the node.

  This tests the ability of the server cluster to fail over all the resources that were owned by the powered down node.

• On one node, unplug the network cables for all networks enabled for intracluster communications.

  This test disables all intracluster network interfaces for the node and prevents the node from exchanging heartbeats with the Cluster Network Driver on the other active nodes. In this event, the Cluster service then initiates failover of the resources owned by the node to the other nodes in the server cluster.

Note

• Disconnecting the SCSI or fiber cable between the host bus adapter and the cluster storage device is not a supported node failure test.

Related Topics