Jackrabbit2.0.0: Why local file system is needed when I just want to use database?

["Niu, Xuetao" <Xu...@fiserv.com>]

Hello,

Would anyone be so kind to tell me why Jackrabbit requires a homeDir on
my file system though I am using database filesystem and database bundle
persistence? 

If I deploy my application in multiple application servers, can they
physically share the same homeDir on a network drive? 

If not, should each homeDir be only accessed by one single JVM? Like,
putting the dir in $user.home$ of each machine?

Is there a way to construct/retrieve jackrabbit repository instance
without a configuration file on disk? How about from an input stream?

Thanks in advance!

Xuetao

Re: Jackrabbit2.0.0: Why local file system is needed when I just want to use database?

[Guo Du <mr...@gmail.com>]

> Would anyone be so kind to tell me why Jackrabbit requires a homeDir on
> my file system though I am using database filesystem and database bundle
> persistence?
There are two type of local storage for default configuration: data and index.
Data is stored in db and some big binary node stored in local file
system (which could configured to store everything in db).
Index is used by lucene which could only be on file system.


> If I deploy my application in multiple application servers, can they
> physically share the same homeDir on a network drive?
You may look for cluster deployment:
http://wiki.apache.org/jackrabbit/Clustering


> If not, should each homeDir be only accessed by one single JVM? Like,
> putting the dir in $user.home$ of each machine?
Data could be only used by single JVM,
Index could be read from multi jvm and write from one jvm.

> Is there a way to construct/retrieve jackrabbit repository instance
> without a configuration file on disk? How about from an input stream?
You could with some coding, see following classes for more details:
org.apache.jackrabbit.core.config.RepositoryConfig
org.apache.jackrabbit.core.TransientRepository(final RepositoryConfig config)

-Guo