so for the past few days I have been playing with contentDB installation. It has been an eye opener. I am relatively new to Oracle (10 months now) and have little experience as a DBA. I have been a Unix admin, software developer, and system architect for a variety of companies. I was a Solaris expert from the early years and remember how difficult it was to install, configure, and upgrade Solaris (at the time called SunOS). I have played with Tomcat at a variety of jobs and even used the iPlanet/Sun Application Server. I am surprised to say that the Oracle suite of products suffers from the same problem that most public domain code and Linux installations suffer from; poor documentation and inadequate dependency checks.
I decided to install contentDB. To do this, I installed the database, version 10.2.0.1 which is the full image that can be downloaded. I got the install working, the os with the right packages, and the accounts and kernel parameters correctly configured. I download and unzip the soa suite and find out that I need to upgrade the database to at least 10.2.0.2. No problem, right? I didn’t initialize any data elements other than the sample database. The upgrade should go easy. WRONG! I downloaded the 800M+ bundle and unzipped it. I execute the runInstaller and watch as the messages go by. I assumed that it would halt the services and upgrade what was needed. It posts a message saying that I need to stop the services managed by the ORACLE_HOME that I have defined. It didn’t say what the processes were or how to see what was running. It just said that I needed to stop something and it was up to me to figure it out. I eventually figured out that I needed to stop the listener, the sql service, the enterprise manager, and the database. It seems like there would be a stopall command that would make this happen.
Once I got everything stopped, I started the upgrade package and it finishes without any warning. Since I only installed the demo tables I thought that I was finished and could connect to the database. WRONG! I forgot to upgrade the database once I upgraded the binary packages. It seems like the upgrade process should have done this for me or at least asked me if I wanted it done. It took me a few hours to figure this one out but I waded through the manuals and forums and got it working.
Now that I have the database upgraded to 10.2.0.3 and OEM reports that it is the right version and everything is properly running. I started the runInstaller for the soa suite. This comes up with a few questions and starts checking for dependencies. I appreciate this but the error message that comes up is of little or no help. Some required prerequisite checks have failed…… What prereqs? A little more detail here would be helpful. A link to a web page? A list of failures?
Am I off base asking for a little more detail here? In the early days of Solaris, these were common error messages. It almost harkens back to Windows messages. “Error: You can’t do anything now” with a Reboot button. Ok, I was a little harsh on the error message, I actually do get an error message in the universal installer screen. It seems like the error window should drop me into the reporting 1 error, 1 requirements to be verified screen that lists the error. The error appears to be easily fixed, gcc-c++ is not found. Fortunately, I had to upgrade the gcc installation for my version of Red Hat Enterprise 4 so I have the iso already mounted. Once I got the prereqs resolved, it complains that there isn’t a ORABPEL schema on the database that I am using. I guess it is back to the manuals to figure out the schema elements that I need to install and configure.