1
Monitoring for Eucalyptus
2
=========================
4
This document explains how to integrate Eucalyptus within a running nagios
5
and ganglia installations. You will then be able to monitor basic status
6
of a running Eucalyptus system. We'll start from Nagios, and you will see
7
ganglia instructions following.
15
Have a Eucalyptus (1.6 or higher) installation running. You should have
16
one CLC and at least one Walrus, SC, CC and NC. For simplicity in this
17
document we expect to install nagios on the CLC machine and that all the
18
machines are 'visible' by the CLC machine.
20
Install nagios on the CLC machine (for example for ubuntu/debian install
21
nagios3) and read the package instructions. You should be able to login to
22
the nagios graphical interface (usually http://localhost/nagios3) and see
23
your host informations there.
28
We will use the passive cheks capability of nagios, and to do so you need
29
to check that the the named pipe exists. Usually is
30
/var/lib/nagios3/rw/nagios.cmd
31
but you can check in nagios.cmd which file that is supposed to be. To
33
mkfifo /var/lib/nagios3/rw/nagios.cmd
34
and be sure the permission and ownership are right
35
chown nagios.www-data /var/lib/nagios3/rw/nagios.cmd
36
chmod 640 /var/lib/nagios3/rw/nagios.cmd
38
To create the configuration for the eucalyptus machine, you can use the
39
script nagios.sh. For example
40
nagios.cmd -setup -nodes "two three four" -cc "one" -cloud one -walrus "one" > eucalyptus.cfg
41
assumes that hosts 'one' has the cloud manager, walrus and the cc, while
42
machines two, three and four are the nc. Inspect the generated file
43
(eucalyptus.cfg) and it all looks good, you can move it to the nagios
44
configuration directory (usually /etc/nagios3/conf.d) and restart nagios.
49
Nagios should now display the Eucalyptus machines and services, but there
50
won't be any data reporting yet. To do so you can now run the same
51
script, in the example
52
nagios.cmd -check -nodes "two three four" -cc "one" -cloud one -walrus "one"
53
(usually the script runs in a cron job). Give sometime to nagios to
54
process the information (more so if you just started it) and you should
55
now be able to monitor the status of Eucalyptus hosts and services.
64
Have a working ganglia installation. In particular you should one of more
65
gmond running on the machines you want monitored (nc, sc and walrus), and
66
at least one gmetad to collect all the data. You should be able to see
67
ganglia's graphs. You need to have a working Eucalyptus installation (1.6
73
You need to have eucalyptus and gmond installed on the machine you want to
79
To start pushing data into ganglia you can use the script ganglia.sh. You
80
will need to specify which type of machine you are monnitoring (nc for
81
node controller, sc for storage controller and walrus) and where
82
Eucalyptus is installed. For example
83
ganglia.sh -type nc -d /opt/eucalyptus
85
You may want to run the above command line from a cronjob.