Mailing-List: contact core-user-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: core-user@hadoop.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
Message-ID: <4806291B.5010204@andremartin.de>
Date: Wed, 16 Apr 2008 18:28:11 +0200
From: =?ISO-8859-1?Q?Andr=E9_Martin?= <mail@andremartin.de>
MIME-Version: 1.0
To: core-user@hadoop.apache.org
Subject: Re: How often Name node fails?
References: <C42941D6.2FCA1%michael@powerset.com>
In-Reply-To: <C42941D6.2FCA1%michael@powerset.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit

Hi Cagdas & Michael,
our cluster works fine with no crashes so far even with more than one 
million files - we have 11 datanodes and one namenode...

Cu on the 'net,
                        Bye - bye,

                                   <<<<< Andr� <<<< >>>> �rbnA >>>>>

Michael Bieniosek wrote:
> From my experience, the namenode almost never crashes.  It does use a lot of RAM, but not nearly as much as it used to.  You might have problems if you have many many small files (maybe 100,000s?).
>
> I have had more problems with datanode failures, particularly when there is mapreduce-generated load on the machines running the datanodes.
>
> -Michael
>
> On 4/14/08 3:14 PM, "Cagdas Gerede" <cagdas.gerede@gmail.com> wrote:
>
> From your experience with Hadoop Distributed File System,
> how reliable is the namenode? How often does it fail? Are there mechanisms
> developed outside of Hadoop to make namenode more fault tolerant?
>
> Thanks for your feedback,