This gem provides a singular, status-checking endpoint to your application. The endpoint provides a JSON response with either an HTTP 200 or an HTTP 500 response, depending on the current application environment.
When the environment tests successfully, an HTTP 200 response is returned with the current application Git revision and server time:
{
"revision": "906731e6467ea381ba5bc70f103b85ed4178fee7",
"time": "2013-05-19T05:38:46Z"
}
When an application validation fails, an HTTP 500 response is returned with a collection of error messages, similar to the Rails < 4.2 responders for model validations:
{
"errors": {
"database": ["The application database is inaccessible or unavailable"]
}
}
To install, add this line to your application's Gemfile:
gem 'rapporteur'
And then execute:
$ bundle install
Supported Ruby versions:
- MRI 2.1
- MRI 2.0.0
- MRI 1.9.3
- MRI 1.9.2
- JRuby 1.7
Supported Rails versions:
- Rails 4.x.
- Rails 3.2.x.
Non-Rails environments are also supported (such as Sinatra), see Usage below for details.
By default, there are no application checks that run and the status endpoint simply reports the current application revision and time. This is useful for a basic connectivity check to be watched by a third party service like Pingdom for a very simple, non-critical application.
You may optionally use any of the pre-defined checks (such as the ActiveRecord connection check) to expand the robustness of the status checks. Adding a check will execute that check each time the status endpoint is requested, so be somewhat wary of doing too much. See more in the Adding checks section, below.
Further, you can define your own checks which could be custom to your
application or environment and report their own, unique errors. The only
requirement is that the check objects are callable (respond to #call
, like a
Proc). See more in the Creating custom checks
section, below.
This gem provides a new, single endpoint in your application. Specifically, it
creates a named /status.json
route, with the "status" name. It does not match
on any other format or variation, which isolates the pollution of your
application routes.
If you'd like to link to the status endpoint from within your application (why, I couldn't guess), you can use a standard Rails URL helper:
link_to status_path
Were you already using the /status.json
endpoint or the "status" route name?
Hmm. Well... you just broke it.
If your application does not have Rails loaded, the Rails Engine logic will be bypassed. In this case, you must configure and run your Rapporteur reports manually because, in the very least, your status route will not be automatically generated as previously described.
So, here is an example usage in Sinatra:
require "rapporteur"
require "sinatra/base"
class MyApp < Sinatra::Base
get "/status.json" do
content_type :json
body Rapporteur.run.as_json.to_json
end
end
Require rapporteur
and then, at some point, call Rapporteur.run
to execute
the configured checks. as_json
will convert the result of the run
into a
Hash which is ready for JSON generation. At that point, do whatever you need to
do with the Hash to have your framework of choice generate and respond with
JSON.
This gem ships with the following checks tested and packaged:
- Rapporteur::Checks::ActiveRecordCheck - Performs a trivial test
of the current
ActiveRecord::Base.connection
to ensure basic database connectivity.
To add checks to your application, define the checks you'd like to run in your environment or application configuration files or initializers, such as:
# config/initializers/rapporteur.rb
Rapporteur.add_check(Rapporteur::Checks::ActiveRecordCheck)
Or, make an environment specific check with:
# config/environments/production.rb
MyApplication.configure do
config.to_prepare do
Rapporteur.add_check(Rapporteur::Checks::ActiveRecordCheck)
end
end
It is simple to add a custom check to the status endpoint. All that is required
is that you give the checker an object that is callable. In your object, simply
check for the state of the world that you're interested in, and if you're not
happy with it, add an error to the given checker
instance:
# config/initializers/rapporteur.rb
# Define a simple check as a block:
Rapporteur.add_check do |checker|
checker.add_message(:paid, "too much")
end
# Make and use a reusable Proc or lambda:
my_proc_check = lambda { |checker|
checker.add_error(:luck, :bad) if rand(2) > 0
checker.add_message(:luck, :good)
}
Rapporteur.add_check(my_proc_check)
# Package a check into a Class:
class MyClassCheck
def self.call(checker)
@@counter ||= 0
checker.add_error(:count, :exceeded) if @@counter > 50
end
end
Rapporteur.add_check(MyClassCheck)
Certainly, the definition and registration of the checks do not need to occur within the same file, but you get the idea. Also: Please make your checks more useful than those defined above. ;)
You could create a checker for your active Redis connection, Memcached connections, disk usage percentage, process count, memory usage, or really anything you like. Again, because these checks get executed every time the status endpoint is called, be mindful of the tradeoffs when making a check that may be resource intensive.
If you need to customize the way in which the current application revision is
calculated (by default it runs a git rev-parse HEAD
), you may do so by
modifying the necessary environment file or creating an initializer in your
Rails application:
# config/initializers/rapporteur.rb
Rapporteur::Revision.current = "revision123"
# config/environments/production.rb
MyApplication.configure do
config.to_prepare do
Rapporteur::Revision.current = "revision123"
end
end
You may pass a String or a callable object (Proc) to .current=
and it will be
executed and memoized. Useful examples of this are:
# Read a Capistrano REVISION file
Rapporteur::Revision.current = Rails.root.join("REVISION").read.strip
# Force a particular directory and use Git
Rapporteur::Revision.current = `cd "#{Rails.root}" && git rev-parse HEAD`.strip
# Use an ENV variable (Heroku Bamboo Stack)
Rapporteur::Revision.current = ENV["REVISION"]
# Do some crazy calculation
Rapporteur::Revision.current = lambda { MyRevisionCalculator.execute! }
The success and error messages displayed in the event that application validations pass or fail are all optionally passed through I18n. There are default localization strings provided with the gem, but you may override them as necessary, by simply redefining the proper locale keys in a locale file within your local application.
For example, to override the database check failure message:
# /config/locales/en.yml
en:
rapporteur:
errors:
database:
unavailable: "Something went wrong"
If you created a custom checker which reports the current sky color, for example, and wanted the success messages to be localized, you could do the following:
# /config/initializers/rapporteur.rb
sky_check = lambda { |checker| checker.add_message(:sky, :blue) }
Rapporteur.add_check(sky_check)
# /config/locales/fr.yml
fr:
rapporteur:
messages:
sky:
blue: "bleu"
- Fork it
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Add some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request