There are many scenarios where re-installations may occur on the same box and there will be times we may have to back out of the deployment. At this point, you will run into a situation where the uninstall on Red Hat wasn't clean. In one case, we could not re-install the Enterprise Management on the server that we just uninstalled due to residual files, but we could not find what they were.
We have seen re-installations fail due to the seos.ini configuration file from the previous installation still lingering around. I would recommend checking the system for the seos.ini configuration file before the re-installation - specifically the /etc/seos.ini symbolic link we create during the installation procedure.
Beyond that, I would make sure the /opt/CA/AccessControlServer folder is removed completely. Make sure there is a new JBoss folder deployed between each of the installation attempts. The backend database needs to be dropped and recreated. Any data from a previous installation will cause the new installation to fail as we will not be able to decrypt the database.
I would also check RPM (rpm -qa) for the CAeAC installation package. We use CAeAC to install the endpoint during the Enterprise Manager install. This needs to be removed by using the "rpm -e" or/and "rpm -e --justdb".