NodeBrain looks nice and I know check_multi before but I use the Nagios Addon “Business Process View”.
It has the same abilities and, from my point of view, is much easier to deploy.
See: http://nagiosbp.projects.nagiosforge.org/

It also has a Impact Analys Tool where you can set the state of a service to see the Impacts on your defined processes.

Integrating this processes in Nagios or NagVis is also possible via the bp_cfg2service_cfg.pl that comes with Business Process View.

Cheers,
Khark

Yes you are right, using check_multi would be easier to use in this case.

The purpose of the article was to show how to integrate Nagios with NodeBrain, not be the perfect implementation of the webshop scenario.

Imho the biggest lack in Nagios is that it do not have a rule engine. In most cases it is not necessary but in some cases it is needed. Solutions like check_multi and check_cluster could help a bit. But if you need more advanced rules with for example correlations over time, logs, snmptraps and so on you need a rule engine.

My experience is that management that want a business view of the environment and the people implementing Nagios do not speak with each other. An advanced rule engine could bridge that gap by attract business consultants that normally works with the big four. Solutions like this are seldom a technical problem.

using the Nagios plugin check_multi you could do the whole stuff
pretty easy
check_multi uses perl expressions to do the state evaluation and
is therefore flexible and powerful.

It took me about 5 minutes to write down the sketch of these four services below according to your rules, where there are three services for the server types and one top level service for the webshop itself which ties everything together.

You can find check_multi here:
http://www.my-plugin.de/wiki/projects/check_multi/start

Cheers,
-Matthias

> * Webserver rules
> o If 3 or more webserver works the webservice is OK
> o If 2 webservers works the webservice is WARNING
> o If 1 webserver or less is working the webservice is CRITICAL
> * Applicationserver rules
> o If 1 or 2 application servers works the application layer is OK
> o If zero application servers works the application layer is CRITICAL
> * Database server rules
> o If 2 or more database server works the database layer is OK
> o if 1 database server works the database layer is WARNING
> o If no database servers works the database layer is CRITICAL
> * The webserver layer, application layer and database layer should be viewed seperatly
> * The total webshop status has the highest status value of webserver layer, application layer a

web.cmd:
# call: check_multi -f web.cmd
statusdat [ web1 ] = webserver1:webservice1
statusdat [ web2 ] = webserver2:webservice2
statusdat [ web3 ] = webserver3:webservice3
statusdat [ web4 ] = webserver4:webservice4
statusdat [ web5 ] = webserver5:webservice5
state [ WARNING ] = count(OK)<=2
state [ CRITICAL ] = count(OK)<=1

app.cmd:
# call: check_multi -f app.cmd
statusdat [ app1 ] = appserver1:appservice1
statusdat [ app2 ] = appserver2:appservice2
state [ CRITICAL ] = count(OK)<=1

db.cmd:
# call: check_multi -f db.cmd
statusdat [ db1 ] = dbserver1:dbservice1
statusdat [ db2 ] = dbserver2:dbservice2
state [ WARNING ] = count(OK)<=2
state [ CRITICAL ] = count(OK)<=1

webshop.cmd:
# call: check_multi -f webshop.cmd
statusdat [ web ] = nagiosserver:web
statusdat [ app ] = nagiosserver:app
statusdat [ db ] = nagiosserver:db