This procedure is based on a configuration with four Node Agents in two resource groups, as described in Example of Creating the Failover Node Agent Component in the Sun Cluster HA for Sun Java System Application Server. The resource groups are na-rg1 and na-rg2, the failover IP addresses are IP1 and IP2, and the eight server instances are Instance1 through Instance8. The failover IP addresses are represented by logical host names.
Perform this procedure on all of the potential primary nodes of the Node Agent component. This procedure requires a pair of nodes, indicated as Node1 and Node2.
Become superuser on a cluster member, Node1.
Bring the Node Agent resource group na-rg1 online on Node1.
# scswitch -z -g na-rg1 -h Node1
Bring the Node Agent resource group na-rg2 online on Node1.
# scswitch -z -g na-rg2 -h Node1
Connect to the server instances from a web browser.
http://IP1:Instance1–port http://IP1:Instance2–port http://IP1:Instance3–port http://IP1:Instance4–port http://IP2:Instance5–port http://IP2:Instance6–port http://IP2:Instance7–port http://IP2:Instance8–port
Ensure that you can connect to any application deployed on the server instances.
If you are not able to connect to the instances from the browser, perform the following actions.
Ensure that your installation satisfies the restrictions and requirements in Configuration Restrictions and Requirements.
Ensure that you correctly followed the instructions in How to Install and Configure the Sun Java System Application Server.
Switch the resource groups from Node1 to Node2.
# scswitch -z -g na-rg1 -h Node2 # scswitch -z -g na-rg2 -h Node2
Test high availability by killing the Node Agent processes manually.
When the Node Agent Process is killed, the Node Agent and all its associated server instances are restarted.
When the application server instances are killed, the Node Agent restarts them. The data service monitors the health of the Node Agent only.
Test the data service functionality by forcing a Node Agent resource failover.
Note the values of the standard resource properties Retry_count and Retry_interval. Kill the Node Agent process manually a number of times equal to Retry_count and within the time period Retry_interval. The Node Agent resource fails over to another cluster node.
When the Node Agents start on another node, repeat Step 4 to test the availability of the server instances.