How to monitor a custom thing (a sql query) in Nagios on Centos

Question

I have spent days reading through pages of cryptic and unintelligible documentation, and incomplete or poorly documented examples.

I cant find where to start - its bewildering!

Basially, I just want to run a simple shell script on a target server and have nagios act on the result. I dont want to have to learn "C" and spend months developing a plugin, which is how it is looking.

E.g. I want to say write a script which executes "select count(*) from logged_in_users"

and have nagios graph draw the resultant number (and even alert on it)

I can write a 5 line script on the Database machine which returns the status and info in about 10 seconds. The problem is how to get Naigos to read this value.

I have read a lot of information about plugins, and seen a lot of plugin "scripts", but the information never says where the plugsin run, do they run on the Srever, The client or both, now how they are configured. I dont want the server to have the DB passwords, so I just want to write an abitrary shell script on the target server and have the result monitored.

The target servers already have something called nrpe installed, which I was hoping would allow me to write a simple shell script, but all its "commands" are in binary, and it is completely cryptic, and I cant find any overview documentation.

Can anyone suggest where to start? Has someone writen a "hello world" script? I have not been able to find a google search string which returns any relvant hits. Does noone on the planet want to monitor the results of a simple custom script via nagios?

Even better would be a mysql plugin, which allows you to insert your own bit of sql, but I cant find anything like this.

Thanks!

Answer 1

Yes, you can run nrpe server on the target server. On nagios server, you can use check_nrpe plugin which takes a command as argument. This command should be defined nrpe server on the target server. So, you script will be executed on the target server by nrpe server.

Here is an example:

You can define a command like this in nrpe.cfg file on target server:

command[check_my_query]=/usr/lib/nagios/plugins/check_my_query.sh

You need to write the script /usr/lib/nagios/plugins/check_my_query.sh on the target server. Of course, you can pass any needed parameters as normal.

On nagios server, you define the check command as follows:

define service{
        use                             generic-service
        host_name                       My_Target_Server
        service_description             My_Query
        check_command                   check_nrpe_1arg!check_my_query
        notifications_enabled           1
        }

You need to use the proper name of check_nrpe_1arg you find in your nagios configuration. It can be already defined like:

define command {
        command_name    check_nrpe_1arg
        command_line    /usr/lib/nagios/plugins/check_nrpe -H $HOSTADDRESS$ -c $ARG1$ -t 15
}

Answer 2

It sounds like you really want to be walked through how to use check_nrpe with a custom defined set of parameters applied to an existing plugin.

First install the mysql check plugin on your mysql server. Figure out the command that you want to execute and whatever parameters you need. Write these settings down.

Install the NRPE daemon on your mysql server. Open port 5666 in iptables if necessary. Edit the file /etc/nagios/nrpe.cfg and add a line like the following.

command[annoying_mysql_check]=/usr/lib/nagios/plugins/check_mysql -w 5 -c 10 "select top 10 from .. order by"

From your Nagios server (at the command line)

check_nrpe -H <your mysql host> -c annoying_mysql_check

This will return the results that you found in the very first set of instructions above.

Once you have this working from the command line on your Nagios server convert this to the syntax and configuration files used by Nagios to automate the scheduling of the checks. This is part of the core Nagios documentation.

Answer 3

http://nagiosplugins.org/man/check_mysql_query which is part of the Nagios package on Ubuntu. Not sure if it's in the Nagios package in CentOS (it might be in a different package than the straight Nagios server package).

Checking the output is a matter of putting the check_mysql_query as a "command" in Nagios and having the Nagios server run it. You can execute the command from any machine that has access to the MySQL server, be it the Nagios server or some other machine (in which case you'd set up NRPE on that machine, which the Nagios server would hit to run the command).

Graphing is kind of another matter. You can use nagiosgraph, which I find an absolute pain to set up. There's also Munin.