jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sankeyraut <sankeyr...@gmail.com>
Subject Jackrabbit - Websphere Clustring
Date Wed, 22 Jan 2014 08:42:21 GMT
Hi All,

We have an application based on jackrabbit repository.
We have 3 static repository and user can create more repositories through
user interface.

We have a clustered deployment and following is the topology. We are using
websphere for deploying the web application
Machine 1

	Websphere1
		Profile1
			WebApplication
		Profile2
			WebApplication

Machine 2			
	Websphere2
		Profile1
			WebApplication
		
All three deployment are part of same cluster and hence point to same
database.

Repositories folder are created in profiles directory in respective
profiles. 
	TransientRepository(File xml, File dir)
	Eg : TransientRepository(new File("repository.xml"), new
File("userrepo1"));

The problem is we need to create the repositories folder outside the
application server installation directory.

If the application is deployed on seperate machine all works fine. But in
case where there are 2 profiles in same websphere on single, it becomes a
problem
	Eg: on profile 1 :  TransientRepository(new File("repository.xml"), new
File("/data/userrepo1"));
		on profile 2 :  TransientRepository(new File("repository.xml"), new
File("/data/userrepo1"));
		
One probable soultion which i can think of was to change user.dir system
parameter to /data. This would help TransientRepository(new
File("repository.xml"), new File("userrepo1")) create 
user repository in /data directory outside the websphere. We can have
seperate system property for user.dir in both the profiles of websphere. But
since there are multiple applications 
deployed on websphere profile, its avoidable.

Other solution seems to extend
org.apache.jackrabbit.core.util.db.ConnectionHelper and override
prepareDbIdentifier method. But it has been decalred as final.
It would be really helpful if we remove the final constrain , we can easily
override it and create a logic to implement our own identifier as per our
need.

Also, is there any other option or workaround that anyone can think of?

Peace.
Sanket Raut



--
View this message in context: http://jackrabbit.510166.n4.nabble.com/Jackrabbit-Websphere-Clustring-tp4660180.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.

Mime
View raw message