Clustering the Application Server
Tomcat 6 supports clustering of two or more application servers for session replication and failover. And because BI platform sessions are serialized, a user session can fail-over seamlessly to another instance of Tomcat, even when an application server fails. For example, if a user is connected to an application server that fails while the user is navigating a folder hierarchy. With a correctly configured application server cluster, the user may continue navigating the folder hierarchy without being redirected to the login page or to the root folder.
For more information about configuring Tomcat 6 in a cluster, see the "How To" here http://tomcat.apache.org/tomcat-6.0-doc/cluster-howto.html.
To configure the Application Server cluster
- Log in to machine
vantgvmwinpb02using the account
- Make a backup copy of the
server.xmlfile located in
c:\tomcat6\conf, renaming it (for example)
- In a text editor, open
server.xml, go to the cluster section, and replace the following:
With the following:
The code for <cluster> represents the default cluster configuration, which functions properly for this pattern. Cluster membership is defined by the Multicast address 188.8.131.52. That value changes between independent clusters if you have several running on the same network. Session replication takes place over TCP/IP and uses port 4000 for communication.
4. Save and close the file.
5. Repeat Steps 1 - 4 on the machine named
6. When you have completed setting up clustering on
vantgvmwinpb03, restart Tomcat.
The output from
catalina.out will have an additional component when the second cluster member starts. At that point, the member will join the cluster. This is indicated in Tomcat as follows:
The Tomcat cluster is now functional.