Creating clusters

Why and when to perform this task

You can manage application servers collectively using a cluster. Use the Server Cluster page to view and manage the cluster.

Note: If you plan to create a cluster of servers that will span multiple systems in a sysplex and will have stateful session beans with an activation policy of Transaction deployed in them, the passivation directory should reside on an HFS (hierarchical file system) that is shared across the multiple systems in the sysplex on which the clustered servers will be running. For more information, see Considerations for clustered servers and stateful session beans.

Steps for this task

  1. Go to the Server Cluster page. Click Servers > Clusters. The Server Cluster page lists clusters of application servers in the cell and states whether a cluster is stopped, started or unavailable.
  2. Click New to access the Create New Cluster page.
  3. Type a cluster name.
    The name must be unique among clusters within the containing cell.
  4. To enable or disable node scoped routing optimization, select Prefer local enabled. The default is enabled, which indicates that, if possible, EJB requests are routed to the client's node. If you enable this feature, performance is improved because client requests are sent to local EJBs.
  5. To enable memory-to-memory replication of HttpSession (for failover) or replication of cached data and cache invalidations with a Web Container's dynamic caching, select Create a replication domain for this cluster. For more information about replication, see Replicating data.
  6. Choose whether to create an empty cluster or to create a cluster based on an existing server.

    To create an empty cluster, do not include an existing server in this cluster.

    To create a cluster based on an existing server, choose Select an existing server to add to this cluster and select the server you want to add. The server you add becomes a template for any additional cluster members that you add to the cluster. Be sure that the template is configured correctly before adding more cluster members that are based on this template.

  7. Click Next.
  8. Add application servers (cluster members) to the cluster. For each new cluster member, do the following:
    1. Type the name of a new application server (cluster member) to add to the cluster.
      On most platforms, the name of the server is the process name. The name must match the (object) name of the application server and must be unique within the cell. Two cluster members cannot have the same name even if they are on different nodes.
    2. Select the node on which the server will reside.
    3. Make sure Generate Unique HTTP Port enabled.
    4. Specify whether to create a replication entry for the server. A replication entry enables memory-to-memory replication of HttpSession (for failover) or replication of cached data and cache invalidations with a Web Container's dynamic caching.
    5. Click Apply to finish the cluster member. Repeat the above steps to define another cluster member.
  9. Click Next and review the summary of changes.
  10. Click Finish to complete the configuration.
  11. Define a Configuring virtual hosts with a unique port number.
    When you define a virtual host with a unique port for this cluster, you guarantee that members of this cluster are not attempting to use a port that is already assigned to another server that is defined on the same node as the cluster.
    1. Go to the Virtual Hosts page of the administrative console.
    2. Click Environment > Virtual Hosts from the navigation tree of the administrative console.
    3. Click New and, on the settings page for a virtual host that displays, specify an administrative name for the virtual host.
    4. Under Additional Properties, click Host Aliases and define an unique port number for the virtual host.

      Note: On z/OS, defining virtual hosts is optional.

  12. Associate this virtual host and port number with each cluster member.
    This enables each cluster member to know which port to listen on for requests from the virtual host.
    1. Go to the Application Servers page of the administrative console.
    2. Select one of the application servers that is a member of your cluster.
    3. Under Additional Properties, click Web Container.
    4. On the Web Container page, under Additional Properties, click HTTP Transport.
    5. Click New and, on the settings page for an HTTP Transport, specify the virtual host name and port number defined in the previous step.
    6. Specify the virtual host name and port number for each cluster member by repeating the previous steps.

    Note: On z/OS, defining virtual hosts is optional.

  13. Click Save on the administrative console taskbar and save your administrative configuration. As part of saving the change to the configuration, you can select Synchronize changes with Nodes before clicking Save on the Save page.
  14. Before you can start the cluster, the configuration needs to be synchronized to the nodes. If you selected Synchronize changes with Nodes when saving your configuration in the previous step, you can ignore this step. If you are running automatic synchronization, wait until synchronization runs. Or, run manual synchronization to get the configuration files moved to the nodes. Click System Administration > Nodes and, on the Nodes page, select the node and click Synchronize or Full Resynchronize. The Nodes page displays status indicating whether the node is synchronized.
  15. To further configure a cluster, click Servers > Clusters. To view the settings for the cluster, click on the cluster Name. Note that unless you have clicked Save and saved your administrative configuration, you only see the Configuration and Local Topology tabs; to see the Runtime tab as well you must save your administrative configuration. Also, ensure that changes are synchronized to the nodes (step 12).

What to do next

If you have made changes to plug-ins, transports, or virtual hosts you must regenerate the plug-in configuration. See Regenerating Web server plug-in configurations for more information. For more information about setting up clustering, see Balancing workloads with clusters.

Related tasks
Balancing workloads with clusters
Related reference
Server cluster collection
Considerations for clustered servers and stateful session beans



Searchable topic ID:   trun_wlm_cluster
Last updated: Jun 21, 2007 9:56:50 PM CDT    WebSphere Application Server for z/OS, Version 5.0.2
http://publib.boulder.ibm.com/infocenter/wasinfo/index.jsp?topic=/com.ibm.websphere.zseries.doc/info/zseries/ae/trun_wlm_cluster.html

Library | Support | Terms of Use | Feedback