Mailing-List: contact solr-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: solr-user@lucene.apache.org
Received-SPF: pass (nike.apache.org: domain of erickerickson@gmail.com
 designates 209.85.128.176 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <4C631781-1F92-4A9E-9E10-1048675386E5@coupang.com>
References: <4C631781-1F92-4A9E-9E10-1048675386E5@coupang.com>
Date: Tue, 24 Jun 2014 08:17:46 -0700
Message-ID: 
 <CAN4YXvcLjM7OY004xjxPsk4Oh=h3iQ6NVTz49Nn8iFYt6Z171g@mail.gmail.com>
Subject: Re: RAMDirectoryFactory setting on replication slave
From: Erick Erickson <erickerickson@gmail.com>
To: solr-user@lucene.apache.org
Content-Type: text/plain; charset=UTF-8

Please don't. At least not until you prove that this
is where your bottleneck is. You haven't described
what you're trying to fix by making such a change.

Solr/Lucene already works a _lot_ to keep the relevant bits of
the index in memory. Additionally, the defaults use
MMapDirectory, which makes use of the OS cache
to hold yet more of the index in memory, see:
http://blog.thetaphi.de/2012/07/use-lucenes-mmapdirectory-on-64bit.html

This feels like an "XY" problem, what is the reason you're
interested?

Best,
Erick

On Tue, Jun 24, 2014 at 1:25 AM, Lee Chunki <lck7141@coupang.com> wrote:
> Hi Guys,
>
> As I know RAMDirectoryFactory setting does not work with replication.
> ( https://cwiki.apache.org/confluence/display/solr/DataDir+and+DirectoryFactory+in+SolrConfig )
>
> By the way,  can I use it for replication slave nodes ( not master )
> or for SolrCloud ?
>
> Thanks,
> Chunki.
>