Bundling dependencies
=====================

Bundling cucumber-nagios's dependencies allows you to drop your cucumber-nagios
project to any machine and have it run. This can be useful if you want to
develop your tests on one machine, and deploy them to another (like a production
Nagios server).

You'll need to bundle your dependencies to use cucumber-nagios.

First you need to make sure the following dependencies are installed:

  - RubyGems
  - bundler gem (automatically pulled in by the cucumber-nagios gem)

To bundle your dependencies, within your project directory run:

    $ gem bundle


Version control
===============

It's strongly recommend that you store your cucumber-nagios projects in a
version control system!

To get up and running with git:

    $ git init
    $ git add .
    $ git commit -m 'created cucumber-nagios project'

To get up and running with bzr:

    $ bzr init
    $ bzr add
    $ bzr commit -m 'created cucumber-nagios project'

.bzrignore and .gitignores are created when you generate a project.


Writing features
================

You can use the cucumber-nagios-gen command to generate new features for
you. It takes two arguments: the site you're testing, and feature you're testing:

    cucumber-nagios-gen feature gnome.org navigation

This will generate two files:

    features/gnome.org/navigation.feature
    features/gnome.org/steps/navigation_steps.rb


As for writing features, you'll want to have a read of the Cucumber
documentation[0], however your tests will look something like this:

    Feature: google.com.au
      To broaden their knowledge
      A user should be able
      To search for things

      Scenario: Searching for things
        When I go to "http://www.google.com"
        When I fill in "q" with "wikipedia"
        And I press "Google Search"
        Then I should see "www.wikipedia.org"


There's a collection of steps that will cover most of the things you'll be
testing for in features/steps/webrat_steps.rb.

You can write custom steps for testing specific output and behaviour, e.g.
in features/smh.com.au/smh.feature:

    Feature: smh.com.au
      It should be up
      And provide links to content

      Scenario: Visiting home page
        When I go to http://smh.com.au/
        Then I should see site navigation
        And there should be a section named "Opinion"

There aren't steps for "Then I should see site navigation", so you have to
write one yourself. :-) In features/smh.com.au/steps/smh_steps.rb:

    Then /^I should see site navigation$/ do
      doc = Nokogiri::HTML(response.body.to_s)
      doc.css("ul#nav li a").size.should > 5
    end

You can use Nokogiri for testing responses with XPath matchers and CSS
selectors.

I suggest you use `cucumber` directly so you can get better feedback when
writing your tests:

    cucumber --require features/ features/smh/smh.feature

This will output using the default 'pretty' formatter.

Running
=======

Invoke the Cucumber feature with the cucumber-nagios script:

    bin/cucumber-nagios features/smh.com.au/smh.feature

cucumber-nagios can be run from anywhere:

    /path/to/bin/cucumber-nagios /path/to/features/smh/smh.feature

It should return a standard Nagios-formatted response string:

    Critical: 0, Warning: 0, 2 okay | passed=2, failed=0, total=2

Steps that fail will show up in the "Critical" total, and steps that pass
show up in the "okay" total.

The value printed at the end is in Nagios's Performance Data format, so it
can be graphed and the like.


Quirks & Caveats
================

Multiple scenarios
------------------

You may want to think about keeping to one scenario to a file, otherwise
you'll get multiple lines of output for a test:

    Critical: 1, Warning: 0, 2 okay | passed=2, failed=1, total=3
    Critical: 1, Warning: 0, 4 okay | passed=4, failed=1, total=5

That said, Nagios should only read the last line, so this might be an ok
behaviour when you want to test for an aggregate of failures across a site.


Failure *is* an option (exceptions are good)
--------------------------------------------

Exceptions raised within your tests will appear in the failed totals, so you
don't need to worry about trying to catch them in your own custom steps.

i.e. if you try fetching a page on a server that is down, or the page returns
a 404, the exception raised by Mechanize just gets treated by Cucumber as a
test failure.


Deploying to production
=======================

Once you've copied your project around, just run the bundler again:

    $ gem bundle

You'll need to have RubyGems and the bundler gem installed on the system
you're deploying too. I know, this is not optimal, but hopefully the bundler
gem will handle this better in the future.


[0] http://wiki.github.com/aslakhellesoy/cucumber