monitoring ubuntu web servers with nagios3
I have chosen Nagios to keep track of the anyhosting.com network. There are many alternatives (some I have explored and some not yet), what I like about Nagios:
- I’ve been using it for a long time; familiarity
- very simple/powerful plugin system
- tons of users, so lots of examples and plugins already available
Nagios version 3 is provided in the Ubuntu repositories, and is quite simple to install:
The default config comes set up to monitor a set of services on localhost; I don’t really like the default Ubuntu/Debian setup of having one config file per host/service/etc, so on the master I’ve replaced the config file structure:
root@admin:/etc/nagios3/conf.d# ls
contacts.cfg extinfo.cfg groups.cfg hosts.cfg services.cfg
timeperiods.cfg
groups.conf contains the set of server types that I care about:
define hostgroup {
hostgroup_name http-servers
alias HTTP servers
members localhost
}
# A list of your mysql servers
define hostgroup {
hostgroup_name mysql-servers
alias MySQL servers
}
# A list of your VHosts
define hostgroup {
hostgroup_name http-vhosts
alias Virtual Host HTTP servers
}
Note that the “http-servers” can define “members” (localhost in this case), however in general I do not add members in this file but instead in the hosts.cfg:
host_name anyhosting1
address 1.2.3.4
use generic-host
hostgroups http-servers
}
define host {
host_name example.com
address 1.2.3.4
use generic-host
hostgroups http-vhosts
}
Note the “hostgroups” line; anyhosting1 is the physical server (this monitor is really checking on the reverse proxy), and example.com is a vhost (which is really proxying to a user running Apache for the “example.com” domain). These two checks make sure that the whole system is working and proxying correctly.
Finally, services.cfg brings it all together by defining which groups should run which services:
define service {
hostgroup_name http-servers
service_description HTTP
check_command check_http
use generic-service
notification_interval 0 # set > 0 if you want to be renotified
}
define service {
hostgroup_name http-vhosts
service_description Virtual Host HTTP
check_command check_httpname
use generic-service
notification_interval 0 # set > 0 if you want to be renotified
}
The Ubuntu nagios-plugins package (which by default is installed along with the nagios3 package) contains plugins that can intelligently check MySQL databases, disk space, load average, etc. By default these only work on the local machine, but these can be made to run on remote machines by installing the nagios-nrpe-server package. I will cover this further in a future blog post.
