Need pointers for Jackrabbit database persistence related informations!

[Ashok <as...@gmail.com>]

Hi,

We are using JackRabbit 2.0.

We are currently testing Jackrabbit in clustering setup. We have 2 servers
with Jackrabbit installed and we are storing all the datas in 1 database. I
have attached the repository.xml we are using.

We observed Jackrabbit creating 15 tables for storing workspace, versioning
and persistence manager informations as below:

*Clustering related tables:*

{prefix}_global_revision
{prefix}_journal
{prefix}_local_revisions


*DBFile System related table:*

{prefix}_fsentry


*DbDatastore related table:*

{prefix}_datastore


*Versioning related tables:*

{prefix}_fsentry
{prefix}_binval
{prefix}_bundle
{prefix}_names
{prefix}_refs


*Workspace related tables:*

{prefix}_fsentry
{prefix}_binval
{prefix}_bundle
{prefix}_names
{prefix}_refs


*We want to know in which table actual data is getting stored.* Please also
point us, if there is any description on purpose of each table.

Also we observed *{Clustering_prefix}_journal and {Workspace_prefix}_bundle
table are growing very significantly*. Is there any documented way for
purging these tables? We saw some responses for purging of
{Clustering_prefix}_journal tables stating

 "old journal entries may be removed as follows: after having verified that
every node in the cluster has read all  outstanding changes"

How can we verify that every node in the cluster has read all outstanding
changes programatically? We went through
"DatabaseJournal.RevisionTableJanitor" cleanUpOldRevisions(). Can we use
this? If so, how can we purge {Workspace_prefix}_bundle table?

Any pointers will be of great help.

-- 
Cheers,
Ashok
+91-9600117508

Re: Need pointers for Jackrabbit database persistence related informations!

[Alexander Klimetschek <ak...@adobe.com>]

On 08.05.11 23:48, "Ashok" <as...@gmail.com>> wrote:

We want to know in which table actual data is getting stored.

Most repository data (nodes & properties) should end up in the *_bundle table of the respective workspace (bundle meaning "node bundle = node + its properties" here). Binary properties (such as those inside nt:file nodes) would go into the *_binval or into the datastore, depending on the datastore and binary blob configuration.

Regards,
Alex

--
Alexander Klimetschek
Developer // Adobe (Day) // Berlin - Basel