In JBoss Server Cluster environment, the Manager for JAVA (EJB) can submit the EJB on another JBoss cluster node.
Scenario is the following :
- The Manager for Java (EJB) has been configured to submit the EJB in using the first instance of the JBoss Cluster
(inside the file 'jndi.xml').
- At the first launch of a Uproc, this uproc ends in Completed status (normal behaviour)
- Stop the first instance of the JBoss Cluster
- At the second launch of the same uproc, this uproc ends in Aborted status (normal behaviour)
- During the third launch of the same uproc, this uproc ends in Completed status despite of the fact that the first instanceof the JBoss Cluster is down. The EJB is executed in the second instance of the JBoss Cluster.
Cause type: Other
Root Cause: It's a JBoss Server behaviour caused by the "Automatic Discovery of HA-JNDI Servers" feature.
1. Stop the Manager for JAVA,
2. In the sub-directory ../dumgrejb20/mgr4java_server/lib, create a file 'jndi.properties' which contains the line : jnp.disableDiscovery=true
3. In the $JAVA_HOME/lib (for host where DUAS & Manager for Java - EJB are installed) :
a.) Create a file named jndi.properties containing the line jnp.disableDiscovery=true
b.) Store this file in the directory JAVA_HOME/lib
4. Restart the Manager for JAVA