Monitoring Data - Big Data Made Easy: A Working Guide to the Complete Hadoop Toolset - page 207

Installing Ganglia
To install Ganglia on Centos Linux, you first install the Epel repository so that you have a safe location for the Linux
yum command to source the software from. Be sure to execute the Epel repository steps on all Hadoop cluster nodes
to support the Ganglia install. For my example installation, I enable the Epel repository for Centos 6.x (on each node)
working in /tmp/epel:
[root@hc1nn ~]# cd /tmp ; mkdir epel ; cd epel
[root@hc1nn epel]# wget http://dl.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm
[root@hc1nn epel]# wget http://rpms.famillecollet.com/enterprise/remi-release-6.rpm
[root@hc1nn epel]# rpm -Uvh remi-release-6*.rpm epel-release-6*.rpm
When I check the downloaded files in /etc/yum.repos.d/, I see that four files have been sourced:
[root@hc1nn epel]# ls -1 /etc/yum.repos.d/epel* /etc/yum.repos.d/remi.repo
/etc/yum.repos.d/epel-apache-maven.repo
/etc/yum.repos.d/epel.repo
/etc/yum.repos.d/epel-testing.repo
/etc/yum.repos.d/remi.repo
Next, I enable the remi repository by editing the remi.repo file and setting Enabled to 1 in the [remi] section:
[root@hc1nn epel]# vi /etc/yum.repos.d/remi.repo
[remi]
name=Les RPM de remi pour Enterprise Linux 6 - $basearch
#baseurl= http://rpms.famillecollet.com/enterprise/6/remi/$basearch/
mirrorlist= http://rpms.famillecollet.com/enterprise/6/remi/mirror
enabled=1
Now, I am ready to install the Ganglia software on the Hadoop name node (in my example, this is hc1nn) and all
of the data nodes. I install the following packages on the Name Node server hc1nn, using the Linux yum command as
the root user:
yum install ganglia
yum install ganglia-gmetad
yum install ganglia-web
yum install ganglia-gmond
On the data nodes (hc1r1m1, hc1r1m2, hc1r1m3 ), I install the following components:
yum install ganglia
yum install ganglia-gmond
The Ganglia gmond processes will collect data and pass it to the gmetad process on hc1nn. I can then view the
data via the Ganglia web component. I must, however, tell Ganglia the frequency at which to collect the data. On
hc1nn, I specify the Ganglia data-collection frequency in the file gmetad.conf; in my example, I have set it to be two
minutes (120 seconds):
vi /etc/ganglia/gmetad.conf
data_source "my cluster" 120 hc1nn