Lately, my supervisor asked me to try out Open Monitoring Distribution. I should gather some information and test if it’s suitable for application monitoring.
After searching the web for a nice guide, i noticed that there’s no really clean and simple step-by-step guide for installing it.
So here it is… at first, import the gpg key:
gpg --keyserver keys.gnupg.net --recv-keys F8C1CA08A57B9ED7
gpg --armor --export F8C1CA08A57B9ED7 | apt-key add -
Now install the Open Monitoring Distribution. Simply insert the path to your sources and use apt.
deb http://labs.consol.de/OMD/ubuntu oneiric main' >> /etc/apt/sources.list
sudo apt search omd
sudo apt-get install omd
After this we need to create a so-called site (a site is a single monitoring instance).
omd create "name-of-the-site" //here I'll use nagios1 for name-of-the-site
This site is now empty and therefore nothing is monitored. To add hosts or services, you have to switch the user with:
sudo su nagios1 //remember that nagios1 is the name of the site
This step is important because OMD creates a new user for each site. It avoids the neccessity of root access when administer the site. To change the password of the default user (omdadmin) type:
htpasswd ~/etc/htpasswd omdadmin
Now, as an exampe, we want to monitor the current workstation (resp. the OMD-Server) with OMD to get a short overview how adding hosts works. For this, the following two copy operations are neccessary. Note that OMD is always installed at /opt/omd to keep the managment of different sites simple.
sudo cp /opt/omd/versions/0.56/share/check_mk/agents/check_mk_agent.linux
sudo cp /opt/omd/versions/0.56/share/check_mk/agents/xinetd.conf
The first operations copies the agent (the agent runs on the host). The second operation configures the according xinetd-service.
After restarting xinetd with
sudo service xinetd restart we can test if OMD is reachable with
telnet localhost 6556
and then have a look at the webinterface:
Concluding, OMD and with it Nagios is a nice and powerful tool for network, hardware, server and cluster monitoring. However, a major drawback are the huge overhead if you just want some performance stats and simple graphs.