Hi Alex,

So, one should also store binaries in the database? This would almost force us 
to use a high end database such as Oracle, wouldn't it?

A few days ago I asked a similar question and got an answer from Shaun 
Barriball, who kindly answered that he has been using a DB/FS setup, in 
cluster, for 2 years without any issues.

I'd really like to have my binaries on NFS and metadata on DB, in cluster, if 
possible or unless you think it's better otherwise.

Thanks!
Juan

----- Original Message -----
From: "Alexander Klimetschek" <[email protected]>
To: [email protected]
Sent: Friday, July 3, 2009 8:16:37 AM GMT -03:00 Argentina
Subject: Re: About Jackrabbit clustering

On Fri, Jul 3, 2009 at 7:50 AM, skyleaf<[email protected]> wrote:
>        <PersistenceManager
> class="org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager">
>        </PersistenceManager>

> When I run Jackrabbit, there are three lines of log info:
>
> INFO cluster.ClusterNode: not started: namespace operation ignored.
> INFO cluster.ClusterNode: not started: namespace operation ignored.
> INFO cluster.ClusterNode: not started: namespace operation ignored.

Not sure how these messages build up, but note that you need a
persistence manager that allows for clustering. The
BundleFsPersistenceManager is not one of them. You need a central
database that allows concurrent access from the various nodes. See [1]
for more info. Also note that you might need to change your
workspace.xml files as well (or simply drop the workspaces directory
if you start from scratch) [2].

[1] http://wiki.apache.org/jackrabbit/Clustering
[2] http://jackrabbit.apache.org/jackrabbit-configuration.html

Regards,
Alex

-- 
Alexander Klimetschek
[email protected]

Reply via email to