Oracle’s next generation connection pooling solution, Universal Connection Pool, can be a bit tricky to set up. This is especially true when a JNDI data source is to be used-most example don’t assume such a scenario. A lot of information is out there on the net, but no one seems to have given the full picture. During the research for chapter 11 of “Pro Oracle Database 11g RAC on Linux” I learned this the hard way. Since the book has been published, a few minor changes changes have been made to the software I used at the time, and those merit an update. Please note that this article’s emphasis is to get this example running-it is by no means meant to be secure enough for a production release! You need to harden the setup considerably for production, but it serves well for demonstration purposes (only).
I have used a four node 18.104.22.168 RAC system as the source for my data. A 2 node cluster database with service “TAFTEST” runs on nodes 1 and 2. It’s administrator-managed and the service has both nodes set aside as “preferred” nodes. The database nodes run Oracle Enterprise Linux 5.564bit with RAC 22.214.171.124. For the sake of simplicity, I used my Windows laptop to host the Tomcat instance, which is now updated to version 6.0.30. I am using apache Ant to build the application. The current stable ant build is 1.8.2. My JDK is also upgraded to the latest and greatest, version 1.6.0_23. I am using the 32bit 126.96.36.199 client package to supply me with ons.jar, ojdbc6.jar and ucp.jar. Read the rest of this entry »