Persisting SearchIndex to database

["Dube, Shreyas" <sd...@akamai.com>]

Hello,

Background: I’m evaluating JCR/Jackrabbit as a layer of abstraction for persistence over Oracle. We are using Tomcat6 as the application server. We usually have 4 – 8 Tomcat’s clustered up in the production environment. I have configured all <FileSystem> and <PersistenceManager> stanzas to point to Oracle (well, I’m using MySQL now, but that’ll change). I have NOT set up clustering in Jackrabbit yet and that is something I want to look at once I figure <SearchIndex> out. This is my <SearchIndex> configuration for a <Workspace>

        <SearchIndex class="org.apache.jackrabbit.core.query.lucene.SearchIndex">
            <param name="path" value="${wsp.home}/index"/>
            <param name="extractorPoolSize" value="2"/>
            <param name="supportHighlighting" value="true"/>
        </SearchIndex>

This <SearchIndex> configuration ends up creating a bunch of files under ${wsp.home}/index under the repository home and the workspace home directory.
Questions:

 1.  What are these files storing? I’m assuming it is the search index.
 2.  Given that we have a clustered Tomcat setup, would it be possible to persist the search index to the database?
 3.  If 2 is not possible, are there other search providers that would work in a clustered environment?
 4.  Would turning on Jackrabbit clustering sync search indexes across all nodes? When are search indexes updated in a clustered environment?

Thanks,
Shreyas

Re: Persisting SearchIndex to database

[Thomas Mueller <mu...@adobe.com>]

Hi,

If you are evaluating, maybe better try Jackrabbit Oak than Jackrabbit
2.x. Oak is preferred for new projects.

But as for Jackrabbit 2.x, see
http://wiki.apache.org/jackrabbit/Clustering : "However, each cluster node
needs its own (private) repository directory, including repository.xml
file, workspace FileSystem and Search index." The search index is Lucene,
which stores the file in the file system. Jackrabbit 2.x does not have an
option to store it somewhere else. (Jackrabbit Oak does by the way.)

Regards,
Thomas




On 10/04/14 16:02, "Dube, Shreyas" <sd...@akamai.com> wrote:

>Hello,
>
>Background: I¹m evaluating JCR/Jackrabbit as a layer of abstraction for
>persistence over Oracle. We are using Tomcat6 as the application server.
>We usually have 4 ­ 8 Tomcat¹s clustered up in the production
>environment. I have configured all <FileSystem> and <PersistenceManager>
>stanzas to point to Oracle (well, I¹m using MySQL now, but that¹ll
>change). I have NOT set up clustering in Jackrabbit yet and that is
>something I want to look at once I figure <SearchIndex> out. This is my
><SearchIndex> configuration for a <Workspace>
>
>        <SearchIndex
>class="org.apache.jackrabbit.core.query.lucene.SearchIndex">
>            <param name="path" value="${wsp.home}/index"/>
>            <param name="extractorPoolSize" value="2"/>
>            <param name="supportHighlighting" value="true"/>
>        </SearchIndex>
>
>This <SearchIndex> configuration ends up creating a bunch of files under
>${wsp.home}/index under the repository home and the workspace home
>directory.
>Questions:
>
> 1.  What are these files storing? I¹m assuming it is the search index.
> 2.  Given that we have a clustered Tomcat setup, would it be possible to
>persist the search index to the database?
> 3.  If 2 is not possible, are there other search providers that would
>work in a clustered environment?
> 4.  Would turning on Jackrabbit clustering sync search indexes across
>all nodes? When are search indexes updated in a clustered environment?
>
>Thanks,
>Shreyas