When you can install a process application on a cluster in which no servers are running

This topic explains the exceptional circumstances in which you might need to install an application on a cluster that has no running servers.

During the installation of a business process application on a server, the Java Naming and Directory Interface (JNDI) name of the data source of the corresponding business process container must be resolved. You cannot, therefore, install an application without a server connection. In a Network Deployment (ND) environment, this server is the deployment manager.

Restrictions lifted

If you want to install a business process application on a cluster in an ND environment, no server in the cluster need be running if the following conditions are true:
  • The required data sources are defined at the cell level.
  • The process application does not specify human tasks.

For process applications that have no human tasks, the data source lookup operation is accomplished within the namespace of the deployment manager, when a lookup operation in the namespace of the application server previously failed. If the application is successfully installed, ignore any error messages in the SystemOut.log file that indicate a failure of the data source lookup operation within the application server namespace.

When it will work

When it will not work

Process applications that contain human tasks require an additional JNDI name lookup operation to locate the staff plug-in provider. Therefore, to help ensure successful installation of such applications, make sure that the cluster includes a running server.

Scoping side effects

A side effect of the name lookup is that if an application server is not running and a data source is defined on its server or node level with the same name as a data source at the cell level, the cell level data source takes precedence. This means that you are using a different data source during deployment and at run time.

Attention: Avoid name clashes. If you define data sources at the cell level manually, use JNDI names that are scoped by the cluster name or server name and node name, for example, jdbc/BPEDB_.
Related reference
Using the bpeconfig.jacl script to configure Business Process Choreographer

Terms of use | | Broken links

Last updated: Mon Mar 27 18:04:05 2006

(c) Copyright IBM Corporation 2005.
This information center is powered by Eclipse technology (http://www.eclipse.org)