0

I am working in Drools-Guvnor-5.5.0.Final version and have configured repository.xml to save all assets + indexes in oracle db. But its not working like that. I want to deploy guvnor in a resilliancy environment where primary and secondary guvnor nodes are pointing to the same DB instance. I deployed guvnor on primary node with repository.xml pointing to ORACLE11 DB and imported/created some assets. Then deployed secondary guvnor pointing to the same DB. I am not able to see the packages and assets I have in primary Guvnor node. I can see the tables created in DB.

I am attaching the repository.xml configurations. In repository.xml I have configured the persistence manager defined in JackRabbit documentation for persistence of assets in db. While I have also used the auto generated repository file. This I generated from Administration panel in Guvnor.

Guvnor should not be keeping assets in Filesystem. Its required for Failover strategies to be keeping assets in DBs.

Kindly copy config file from the below link, https://community.jboss.org/thread/230419

learner
  • 906
  • 2
  • 10
  • 39
  • 1
    Did you also change the workspace.xml file? Also, did you consider using the database data store (currently you are using the FileDataStore)? – Thomas Mueller Jul 17 '13 at 11:57
  • @Thomas, Thanks for the valuable input. I configured the datastore both in primary and secondary servers, and now I can see my assets in database. But there is one problem, in secondary server I can see the package names correctly but there are no assets displayed in the packages. – learner Jul 21 '13 at 05:22
  • @Thomas, Along with datastore I have also configured the workspace.xml in the following paths.
    jboss-eap-6.0/bin/repository/workspaces/default
    /apps/jboss/jboss-eap-6.0.1/jboss-eap-6.0/bin/repository/workspaces/security
    – learner Jul 21 '13 at 05:36
  • So, when you say 'primary' and 'secondary' I assume you mean a cluster. However, to use clustering, you need to configure it. See the [wiki page about clustering](http://wiki.apache.org/jackrabbit/Clustering). Also, the workspace _filesystems_ are not typically shared (it might work, but I wouldn't recommend it). – Thomas Mueller Jul 21 '13 at 13:47
  • @Thomas, I am in resillient environment in which I have two linux servers. On each Guvnor is deployed which is pointing to a common database. In my environment only one server is active at one time. If my primary server goes down for any reason(Maintenance or crash) then only the secondary server will be started and will become active. Now all the clients will be routed to secondary server by resellience configurations. I hope this clears that I am not using clusters where both servers are active at the same time – learner Jul 31 '13 at 07:12
  • @Thomas, have u checked my configuration file uploaded in the jboss community thread. I have pasted the link in my question. Kindly suggest if my configuration is incorrect. I have also come across a forum which says that if u truncate db tables and restart the guvnor server, one can see the assets appear back in DB. I have tested it the sameway and its correct. Now this means that jackrabbit is still saving assets in filesystem. Kindly correct me if I am wrong in configurations. I want to avoid data on filesystem. I want everything to be saved in DATABASE – learner Jul 31 '13 at 07:18
  • 1
    OK, if you don't use Jackrabbit clustering, then the problem might be that the Lucene index is stored in the file system. You would need to synchronize the Lucene index with the other server. – Thomas Mueller Jul 31 '13 at 09:29
  • 1
    So, can I conclude that Anyhow jackrabbit will be writing to file system. And the repository directory will be created and is compulsory for jackrabbit to work. – learner Aug 01 '13 at 07:20
  • Is this also the case with modeshape ? – learner Aug 01 '13 at 07:21
  • 1
    So, can I conclude that even I have configured DBPersistenceManagers and Datastores for DB,jackrabbit will still be writing some info to file system. And the repository directory will be created and is compulsory for jackrabbit to work. Havig said that I have also commented the tag in repository.xml and the reult was ==> repository directory is still there with no index folder. So jackrabbit need the repository folder in anycase – learner Aug 01 '13 at 07:28

1 Answers1

0

Thanks Thomas for the guidance. i tried out Clustering of jackrabbit in repository.xml and it worked for me. Now I have two member Guvnors in a clustered environment. Both poiting to same DB for fileSystem, persistentManager and DataStore.

Now What I did for testing is brought one cluster node down and make some changes in the other one. Now brought the second node alive and running and could see my latest assets even the changes made by other node was also available.

So this resolves my issue of keeping assets in DB to avoid backingup repository directory. Now I can depend on Database.

learner
  • 906
  • 2
  • 10
  • 39