Mailing-List: contact user-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hadoop.apache.org
Received-SPF: pass (nike.apache.org: domain of bejoy.hadoop@gmail.com
 designates 209.85.210.176 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CANw9MFuhJAG32nKr6uyeZU71zgLDeMnXAVJk5g9YxU=bB7Ug4Q@mail.gmail.com>
References: <006C1185901F10488AB4710EB5EE3C020EEECD54@chnshlmbx11>
	<CANw9MFuhJAG32nKr6uyeZU71zgLDeMnXAVJk5g9YxU=bB7Ug4Q@mail.gmail.com>
Date: Thu, 1 Nov 2012 17:13:21 +0530
Message-ID: 
 <CACD21EPM-aAtkJpvtzaggHBRkjkPgBX3RA3aFDJzXSeX=vWvRw@mail.gmail.com>
Subject: Re: Map Reduce slot
From: Bejoy Ks <bejoy.hadoop@gmail.com>
To: user@hadoop.apache.org
Content-Type: multipart/alternative; boundary=14dae934102daac84604cd6d86fd

--14dae934102daac84604cd6d86fd
Content-Type: text/plain; charset=ISO-8859-1

Hi Udippan

The number of map/reduce tasks slots on your cluster is the sum of all the
slots from your TaskTracker nodes. Based on your node's resource
availability you can even configure that on a per node basis. The slots are
defined on a node level using the following properties
mapred.tasktracker.map.tasks.maximum and
mapred.tasktracker.reduce.tasks.maximum .

As Kapil mentioned the total number of slots across the cluster can be
obtained from JT web UI.

The other 3 properties can be defined on a job level. However in production
clustes the jvm size is marked final to prevent abuses that may lead to
OOMs.

The jvm size of task jvms are defined by 'mapred.child.java.opts' which
defaults to 200 Mb and the jvm reuse is defined by
'mapred.job.reuse.jvm.num.tasks' which defaults to 1 task per jvm.

HTH.

Regards
Bejoy KS

--14dae934102daac84604cd6d86fd
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Hi Udippan<br><br>The number of map/reduce tasks slots on your cluster is t=
he sum of all the slots from your TaskTracker nodes. Based on your node&#39=
;s resource availability you can even configure that on a per node basis. T=
he slots are defined on a node level using the following properties mapred.=
tasktracker.map.tasks.maximum and mapred.tasktracker.reduce.tasks.maximum .=
<br>
<br>As Kapil mentioned the total number of slots across the cluster can be =
obtained from JT web UI.<br><br>The other 3 properties can be defined on a =
job level. However in production clustes the jvm size is marked final to pr=
event abuses that may lead to OOMs.<br>
<br>The jvm size of task jvms are defined by &#39;mapred.child.java.opts=
9; which defaults to 200 Mb and the jvm reuse is defined by &#39;mapred.job=
.reuse.jvm.num.tasks&#39; which defaults to 1 task per jvm.<br><br>HTH.<br>
<br><div class=3D"gmail_quote"><font color=3D"#888888">Regards<br>Bejoy KS<=
br></font></div><br>

--14dae934102daac84604cd6d86fd--