Service Mapping weblogic applications from shared linux host

[Preston_83]

We are starting to service map weblogic applications that run from linux host clusters.  Each physical cluster has multiple weblogic domains that run multiple applications. 

Since these clusters are shared resources, our service maps are picking up databases that are used by other applications and not specifically used by the application we are currently mapping.  My theory is that discovery is picking up connections to these other databases using netstat from the linux host and are not used by this specific application.  We have hundreds of maps to create and do not want to have to go through each of them and remove the databases that are not used by this specific application.

Any suggestions?  I have Traffic based discovery disabled.

[Mateus Oliveira · Accepted solution]

Hi @Preston_83 ,

If you open and debug the specific Pattern creating this connection in the Pattern Designer, you should be able to determine exactly the criteria being used by the Pattern to create these relationships and from there you can edit it to what will make sense to you and your applications.

In our scenario when we did Weblogic mapping, we parsed the domains file to get the exact domain and used this information to create the connections. This could also be a possibility for you.

Please, mark my response as helpful in case it helped you.

[Preston_83]

Mateus,

 Looking over the Oracle JDBC Connectivity pattern, the way it is written will not work with the way we deploy our weblogic environment.  I am going to attempt to create my own connectivity pattern and then disable the OoB pattern. 
If you don’t mind, I have a couple quick questions.  

You stated that you used your domains file to get the information needed.  Are you referring to the DOMAIN_HOME\config directory config.xml file, or the jdbc xml file?  Was the data you needed already in the file, or did you have to add it? 

The connectivity pattern starts off with a temporary variable called $conn_table which looks like it is brought in from the parent pattern called identification for weblogic module.  When I attempt to debug the OoB connectivity pattern, the steps say it can’t find the file.  How do you pass a variable table from one pattern to another in debug mode?

[Mateus Oliveira]

Hi @Preston_83 ,

I believe that the Oracle JDBC Connection is actually a Shared Library which is used by the pattern Weblogic Module. So when you are debugging the pattern itself (Weblogic Module), when you get to the Connection Section for the Oracle JDBC Connection, you will have all the temporary variables from the execution of all steps n your Identification and Extension sections. One good approach for you could be to start debugging from the Identification Section so you can confirm which information you are getting for that variable in the first place to be sure what to expect in the Connection section.

As per your first point, we used the domain_home directory, but I am not 100% sure it would be applicable for what you are trying to do. Considering the troubleshooting you have already done, perhaps developing your own connection would be indeed the right way to go.

I hope this helps you in your quest! 🙂

[Preston_83]

Thank you.  I found that this was using the Oracle JDBC pattern and it was using xml files in a directory that is for the whole domain and not each individual cluster.  I will need to look into a solution to parse these files based on the specific cluster this application map belongs to.