R12.2 Cloning and Ports

Cloning in R12.2 is more complex than in previous releases of the E-Business Suite. The introduction of Weblogic Server is the major reason for this.

An installation of EBS R12.2 now uses two port pools, a run pool and a port pool that for the most part must contain distinct ports. This becomes an issue when more than one server is configured on the same apps tier. Since an instance of oacore_server should be limited to about 200 processes, most companies will need multiple oacore_servers. For many companies, there will be sufficient capacity on the apps node to allow these servers to be added on the existing node without having to add an additional apps tier on another node. However, this is not possible if the port for additional managed server is in the port pool for the patch environment. The current recommendation is to make the patch port pool at least 10 higher than the run port pool. For more information on adding managed servers see MOS Note 1905593.1.

The port issue becomes even more complex with cloning using Rapid Clone (MOS Note 1383621.1). Because adpreclone.pl, packages an installation of the fusion middleware from the source, the production port pools will be required while the target apps tier is configured. This means that you need 4 port pools that are distinct across all the managed server ports.

If you use the same port pool and are using a pairs file (perl adcfgclone.pl component=appsTier pairsfile=<PAIRSFILE> addnode=yes dualfs=yes), you will receive an error message that just says pairs file validation failed.

If you are using a different port pool, but one which does not allow sufficient room for the WLS Managed Servers in both between any of the four port pools, you will find the clone fails during either the RUN or PATCH configuration stages.

My recommendation to avoid this issue is to set the production patch port pool far enough from the run port pool to allow for both production and clone target ports to be in the middle.  Set the target patch port pool to be have the same gap.  For example, in production use port pools of 0 and 20, in the target use 10 and 30.  I also strongly suggest that the first time you configure a target, you run rapid clone by hand to generate the pairs file for use in future runs (as opposed to editing a pairs file from another clone).


2 thoughts on “R12.2 Cloning and Ports”

    1. Are asking about a clone or your primary system?

      If a clone, I would rebuild the patch file system.
      If not, well you still need to rebuild the patch filesystem. I would get instructions from support to be sure that you are not hung out to dry if it fails. I believe that you can change the ports in the patch filesystem context file and run AutoConfig. In my case, I realized that I had not left enough room during my test upgrades, so I was able to build with sufficient room between the run and patch port pools. This is something to test heavily in a non production environment. Testing needs to include running through a full patch cycle including cutover.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.