Mailing-List: contact dev-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@jackrabbit.apache.org
Received-SPF: pass (athena.apache.org: domain of b.vanderschans@1hippo.com
 designates 64.18.2.16 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <4DCA6AF6.1060903@liip.ch>
References: <4DBE97F9.3030303@liip.ch>
 <BANLkTinZwot0yYs0+U7UsUWDiBBTELBTAA@mail.gmail.com>
 <4DBEA833.3050104@liip.ch> <4DC153C4.5090300@liip.ch>
 <BANLkTimvSZEr9tbP=HOukE0NTPrd3FD6jA@mail.gmail.com>
 <4DCA6AF6.1060903@liip.ch>
From: Bart van der Schans <b.vanderschans@onehippo.com>
Date: Wed, 11 May 2011 13:14:37 +0200
Message-ID: <BANLkTi=+_14vCfch7K=B_3Boq9DL1QwHnA@mail.gmail.com>
Subject: Re: Add more options to make Jackrabbit more failsafe and/or
 scale-out
To: Christian Stocker <christian.stocker@liip.ch>
Cc: dev@jackrabbit.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Wed, May 11, 2011 at 12:54 PM, Christian Stocker
<christian.stocker@liip.ch> wrote:
>
> Hi
>
> Just one thing to avoid missunderstandings.
>
> On 11.05.11 12:01, Bart van der Schans wrote:
>> Hi Christian,
>>
>
>>
>> Does anybody have some pointer why it's trying to write? It seem to
>> shutdown just fine anyway.. It doesn't make much sense to me to create
>> a ReadOnlyDatabaseFileSystem just to avoid this shutdown issue.
>>
>> I think the solution I created is a bit more narrow as it only aims to
>> allow to run JR on a database slave node independently. I think
>> Christian's idea can be expanded to separate all write database
>> operations from read operations in JR, which would allow for scaling
>> out in heavy read dependent environments =A0with multiple db slaves
>> while keeping a single db master for handling all writes.
>
> My patch only seperates the read/writes for the cluster journal
> operations. It doesn't seperate the "common" read/write request handled
> by the PM. This of course only works, if your Jackrabbit instances using
> the slaves only do reads and no PM-writes at all. You should handle the
> seperation in your application (read from the read-only Jackrabbit, but
> write to the write-enabled Jackrabbits), the same as you would do, you
> if you would use a traditional master/slave MySQL Setup (with no
> Jackrabbit or other CR at all)

That's why I said your patch/idea could be "expanded" ;-) If the db
operations where read/write separated, you wouldn't have to worry in
your application about "write-enabled" JRs. Then JR would take care of
doing the writes in the master database and the reads from the slave
database(s).

Regards,
Bart