offline solr indexing

[Charles Federspiel <ch...@gmail.com>]

Solr Users,
Our app servers are setup on read-only filesystems.  Is there a way
to perform indexing from the command line, then copy the index files to the
app-server and use Solr to perform search from inside the servlet container?

If the Solr implementation is bound to http requests, can Solr perform
searches against an index that I create with Lucene?
thank you,
Charles Federspiel

Re: offline solr indexing

[Otis Gospodnetic <ot...@yahoo.com>]

This should be fine.  You won't have to replicate your index, just reopen the searcher when commit is done, that's all.  Index updates and searches can be happening at the same time.

Otis
--
Sematext -- http://sematext.com/ -- Lucene - Solr - Nutch



----- Original Message ----
> From: Charles Federspiel <ch...@gmail.com>
> To: solr-user@lucene.apache.org
> Sent: Saturday, May 2, 2009 12:40:13 PM
> Subject: Re: offline solr indexing
> 
> Thanks. I imagine replication to a slave would require a filesystem writable
> by that slave.
> 
> I think it helps to realize that indexing is really a function of Content
> Management.  After some discussion with coworkers, I've learned that our
> internal CMS server runs within tomcat and shares a filesystem with our
> public app-servers.  So I'm hoping to deploy solr both to the tomcat
> instance (for indexing) and our other instances (for searching) simply
> sharing a Solr home between them.  How bad is this? Does updating and
> commiting the index interrupt search?  It would only affect our internal
> instance, but I still need to know all the effects of this setup.
> 
> On Mon, Apr 27, 2009 at 6:32 PM, Shalin Shekhar Mangar <
> shalinmangar@gmail.com> wrote:
> 
> > On Tue, Apr 28, 2009 at 12:38 AM, Charles Federspiel <
> > charles.federspiel@gmail.com> wrote:
> >
> > > Solr Users,
> > > Our app servers are setup on read-only filesystems.  Is there a way
> > > to perform indexing from the command line, then copy the index files to
> > the
> > > app-server and use Solr to perform search from inside the servlet
> > > container?
> >
> >
> > If the filesystem is read-only, then how can you index at all?
> >
> > But what I think you are describing is the regular master-slave setup that
> > we use. A dedicated master on which writes are performed. Multiple slaves
> > on
> > which searches are performed. The index is replicated to slaves through
> > script or the new java based replication.
> >
> >
> > > If the Solr implementation is bound to http requests, can Solr perform
> > > searches against an index that I create with Lucene?
> > > thank you,
> >
> >
> > It can but it is a little tricky to get the schema and analysis correct
> > between your Lucene writer and Solr searcher.
> >
> > --
> > Regards,
> > Shalin Shekhar Mangar.
> >

Re: offline solr indexing

[Charles Federspiel <ch...@gmail.com>]

Thanks. I imagine replication to a slave would require a filesystem writable
by that slave.

I think it helps to realize that indexing is really a function of Content
Management.  After some discussion with coworkers, I've learned that our
internal CMS server runs within tomcat and shares a filesystem with our
public app-servers.  So I'm hoping to deploy solr both to the tomcat
instance (for indexing) and our other instances (for searching) simply
sharing a Solr home between them.  How bad is this? Does updating and
commiting the index interrupt search?  It would only affect our internal
instance, but I still need to know all the effects of this setup.

On Mon, Apr 27, 2009 at 6:32 PM, Shalin Shekhar Mangar <
shalinmangar@gmail.com> wrote:

> On Tue, Apr 28, 2009 at 12:38 AM, Charles Federspiel <
> charles.federspiel@gmail.com> wrote:
>
> > Solr Users,
> > Our app servers are setup on read-only filesystems.  Is there a way
> > to perform indexing from the command line, then copy the index files to
> the
> > app-server and use Solr to perform search from inside the servlet
> > container?
>
>
> If the filesystem is read-only, then how can you index at all?
>
> But what I think you are describing is the regular master-slave setup that
> we use. A dedicated master on which writes are performed. Multiple slaves
> on
> which searches are performed. The index is replicated to slaves through
> script or the new java based replication.
>
>
> > If the Solr implementation is bound to http requests, can Solr perform
> > searches against an index that I create with Lucene?
> > thank you,
>
>
> It can but it is a little tricky to get the schema and analysis correct
> between your Lucene writer and Solr searcher.
>
> --
> Regards,
> Shalin Shekhar Mangar.
>

Re: offline solr indexing

[Shalin Shekhar Mangar <sh...@gmail.com>]

On Tue, Apr 28, 2009 at 12:38 AM, Charles Federspiel <
charles.federspiel@gmail.com> wrote:

> Solr Users,
> Our app servers are setup on read-only filesystems.  Is there a way
> to perform indexing from the command line, then copy the index files to the
> app-server and use Solr to perform search from inside the servlet
> container?


If the filesystem is read-only, then how can you index at all?

But what I think you are describing is the regular master-slave setup that
we use. A dedicated master on which writes are performed. Multiple slaves on
which searches are performed. The index is replicated to slaves through
script or the new java based replication.


> If the Solr implementation is bound to http requests, can Solr perform
> searches against an index that I create with Lucene?
> thank you,


It can but it is a little tricky to get the schema and analysis correct
between your Lucene writer and Solr searcher.

-- 
Regards,
Shalin Shekhar Mangar.

Re: offline solr indexing

[Amit Nithian <an...@gmail.com>]

Not sure if this helps but could you make this a solr server that is not
accessible by any other means (except internal), perform your index build
using the dataimporthandler and use Solr's replication mechanisms to move
the indices across?
You can issue the HTTP request to rebuild the index from the command line
(i.e. GET ..)

On Mon, Apr 27, 2009 at 12:08 PM, Charles Federspiel <
charles.federspiel@gmail.com> wrote:

> Solr Users,
> Our app servers are setup on read-only filesystems.  Is there a way
> to perform indexing from the command line, then copy the index files to the
> app-server and use Solr to perform search from inside the servlet
> container?
>
> If the Solr implementation is bound to http requests, can Solr perform
> searches against an index that I create with Lucene?
> thank you,
> Charles Federspiel
>