JackRabbit : How to move all filesystem data into DB. Eliminate data/versioning/files on FileSystem

Question

I have a Jackrabbit as standalone server. I have configured pool.OraclePersistenceManager to keep all my data in Oracle11. Data is moved to DB but still the filesystem directory , i.e."repository", is still there on the server filesystem. If I delete the "repository" directory from filesystem then I cannot get my documents. this is wierd.

I moved repo to Oracle DB to take periodic backups and to make it available for multiple jackrabbit instances residing on different server machines and only one server is active at a time. This I have done to prevent down time, if one server goes down the other jackrabbit instance is brought up and connected to the same DB instance.

One more thing If repo data is truncated from the database then on restart of jackrabbit server data is again visible in DB. Is this the correct behaviour. I think if data file rows are deleted from DB then data should not be available in db even after restart of jackrabbit instance.

I want to eliminate data keeping in file system.

It will be great if someone can describe the jackrabbit directory usage and db usage. And how they are related.

Answer 1

I did the clustering thing and now if one node is down I bring the second node up and all my documents/assets are available. BUT STILL could NOT remove the repository directory from filesystem. Atleast now I don't need to worry about it as I am getting my data available in server crash issues

Answer 2

You might take a look in the clustering documentation as most of the configuration/content has to be shared between nodes thus one of the 'easier' way is to store everything in database. It looks like you have to share the configuration, so i'm afraid you would have to set up some NFS sharing for the configuration files.

Answer 3

The repository configuration (the file repository.xml) needs to be in the file system; I don't a way to avoid that. However, this is just an immutable configuration file, so I wonder if this is really a problem. If yes, why?

If you are using a search index, then its files are stored in the local file system. There is no easy way around that, except if you disable the search index.

Answer 4

I managed to move all my data using clusters and this is serving my purpose of a backup. I created two cluster nodes (following the JackRabbit Clustering Document). Then kept one node as offline and the other as online. So when anything goes wrong then my second cluster will come alive and replace the first cluster node. I am getting modifications made by node A in node B.

So I think my backup and recovery is working great. Atleast we can call this a resellient environment.