• Nagios

Nagios

Escape '&'

In, say, a URL. foo?bar=baz&fuzz=woot will break things as the cmd interpreter will choke on the &. You can't escape with a single \, you need two because nagios uses \ to escape !, so to pass the \ through to the shell, you need to type it twice it, a la foo?bar=baz\\&fuzz=woot.

Configuration

• nagios -v /etc/nagios/nagios.cfg shows error messages on startup, use /etc/rc.d/init.d/nagios after

it starts working.

• Fiddled with /etc/httpd/conf.d/nagios.conf so I could access the pages from my computer.

• Set up apache authentication in /etc/httpd/conf.d/nagios.conf

• Set enable_notifications=0 in /etc/nagios/nagios.cfg

• put username/passwd in /etc/nagios/passwd with htpasswd.

• in /etc/nagios/

  • copy *.cfg-sample -> *.cfg

  • make sure all .cfg files are readable by web server

  • edited nagios.cfg to use minimal.cfg only

  • edited cgi.cfg to authorize myself to view everything

• The blurb in the docs about time saving with templates is really helpful
• check_ntp was a little busted at first
• created two contacts, one is my email and one is my pager. There isn't really a way to give one contact an email and a pager and tell the thing to use them at different times.

ePN

Nagios for Fedora comes default with ePN or Embedded Perl Nagios. That means any plugins written in perl will be interpreted not by perl, but by nagios. There are a lot of wierd caviats that this creates including:

• $0 always equals /usr/sbin/p1.pl - near as I can tell, there is no way to find out what the script was called as
• Calling "exit" from a module causes problems under some circumstances. If possible, use die.

Quieting Down Nagios During Host Service

OK, so you're supposed to use nagios's fancy 'planned downtime' stuff, but that includes letting the CGIs have write access, which I don't like. If you're like me, and you don't want pages when you take a system offline, set up a check host command that always says the host is down, and says the host is being maintained.

This way, you'll get one page when nagios realises the system is offline. You won't get repeated onnline/offline pages from rebooting, and even better, you won't get the 8 pages you get when the system is pingable, but not offering any other services.

Here's a command (in checkcommands.cfg in my setup) that will do just that.

define command{
    command_name check-host-planned
    command_line    $USER1$/check_dummy 2 "Down For Service"
}

Upgrading to Nagios 3 and Making CGIs run as Nagios

CentOS yum repos provide Nagios 2 only. For nagios 3, you need to enable another yum repo.

First, set up rpmforge.repo as a yum repo. See http://wiki.centos.org/AdditionalResources/Repositories/RPMForge

• Important note:

includepkgs = nagios

... to only include nagios from rpmforge. Then run yum update and it will update from nagios 2 to nagios 3.

• I had to change the /etc/rc.d/init.d/nagios script so the su -c ... line looked like su -s /bin/sh -c ....

Okay, it's not working. Need to re-assess cgiwrapness

• made new home dir for nagios /home/system/nagios
• moved /usr/share/nagios to /home/system/nagios/htdocs
• moved /usr/lib/nagios/cgi to /home/systen/nagios/htdocs/cgi-bin
• made symlinks from htdocs and cgi-bin to thier original locations to facilitate rpm updates
• had to create /home/system/nagios/cgiwrap and stick copies of cgiwrap in there
• changed Directory auth stuff to protect /home/system/nagios - this makes /home/system/nagios/htdocs and /home/system/nagios/cgiwrap protected by the same auth stuff
• Made a scriptAlias for /nagios/cgiwrap to /home/system/nagios/cgiwrap

• RewriteRule ^/nagios/cgi-bin/(.*) /nagios/cgiwrap/cgiwrap/nagios/$1 [PT]

Here's the apache snippet. This must go inside the appropriate virtualhost.

RewriteRule ^/nagios/cgi-bin/(.*) /nagios/cgiwrap/cgiwrap/nagios/$1 [PT]

ScriptAlias /nagios/cgiwrap/ /home/system/nagios/cgiwrap/

Alias /nagios /home/system/nagios/htdocs

<Directory "/home/system/nagios">
   AllowOverride None
   Order allow,deny
   Allow from all
   AuthName "Nagios Access"
   AuthType Basic
   AuthUserFile /etc/nagios/passwd
   Require valid-user
</Directory>

The following does some but not all of the required steps

set -e

RPMCGI=/usr/lib/nagios/cgi
RPMHTML=/usr/share/nagios
NAGIOS=/home/system/nagios

function mkdiro {
    dir=$1
    if [ -d $dir ] ; then
        return
    fi
    mkdir $dir
}

mkdiro $NAGIOS

mv $RPMHTML $NAGIOS/htdocs
mv $RPMCGI $NAGIOS/htdocs/cgi-bin

ln -s $NAGIOS/htdocs $RPMHTML
ln -s $NAGIOS/htdocs/cgi-bin $RPMCGI

usermod -d /home/system/nagios nagios
mkdiro $NAGIOS/checkresults
mkdiro $NAGIOS/rw


chown -R nagios:nagios $NAGIOS
chown -R nagios:nagios /var/log/nagios

mkdiro $NAGIOS/cgiwrap
cp /home/webmaster/cgi-bin/cgiwrap* $NAGIOS/cgiwrap

chmod -R a+rX $NAGIOS
chmod -R og-rwx $NAGIOS/checkresults
chmod -R og-rwx $NAGIOS/rw

echo all loadmodule stuff should be done in /etc/httpd/conf/loadmodule.conf
echo Hint! remove LoadModule calls from conf.d/*.conf with a command like this:
echo 'perl -p -i -e \'s/^LoadModule/#LoadModule/\' *.conf'

CategoryNotes