Outpost
Features
Outpost is a tool to monitor the state of your service (not server). What does it mean?
It means:
- it can monitor the state of a server, such as MySQL;
- it can monitor some business rule to see if everything is running accordingly (such as cron jobs)
- it can monitor several servers
- it can monitor whatever you can code with Ruby
It will connect to the related machines (it won't have any proxies/agents running on the servers to report data) and collect the data. The idea is to be completely uncoupled with the systems. It should report a status per declared system.
The idea is to make a reliable framework for the Ruby developer to create his own monitoring rules. So, summing it all up, Nagios in Ruby, much cooler!
Information and examples
Installing
Outpost is tested with Ruby 1.8.7 and Ruby 1.9.3.
gem install outpost
Starting
To create your Outposts, you must require 'outpost'. You also need to include 'outpost/scouts' if you want to use the supplied scouts. Example:
require 'outpost'
require 'outpost/scouts'
class Monitor < Outpost::Application
using Outpost::Scouts::Http => "web page" do
options :host => 'localhost', :port => 3000
report :up, :response_code => 200
end
end
monitor = Monitor.new
monitor.run
p monitor.messages # => ["Outpost::Scouts::Http: 'web page' is reporting up."]
How it works
Consider the following example:
require 'outpost'
require 'outpost/scouts'
class HttpOutpostExample < Outpost::Application
using Outpost::Scouts::Http => "web page" do
options :host => 'localhost', :port => 3000
report :up, :response_code => 200
report :down, :response_body => {:match => /Ops/}
end
end
outpost = HttpOutpostExample.new
outpost.run # => :down
In this simple example, an Outpost was created to monitor a web server running on localhost at port 3000. Every time #run is called, the outpost will run associated rules (in this example, check if the HTTP response code is 200 and report "up" if it does and also check if the response body matches /Ops/, reporting "down" in that case).
Outpost
Outpost is the description of the system and provides a DSL to do it. Check "How it works" section for an example, or check the integration tests for more.
Scout
Scout are pure Ruby classes that will test your server. For instance, check the Outpost::Scouts::Http example below:
module Outpost
module Scouts
class Http < Outpost::Scout
extend Outpost::Expectations::ResponseCode
extend Outpost::Expectations::ResponseBody
attr_reader :response_code, :response_body
def setup(options)
@host = options[:host]
@port = options[:port] || 80
@path = options[:path] || '/'
end
def execute
response = Net::HTTP.get_response(@host, @path, @port)
@response_code = response.code.to_i
@response_body = response.body
end
end
end
end
It must implement the #setup and #execute methods. The magic lies in the #execute method, where you can implement any kind of logic to test whether your system is up or not. You may also include expectations in order to process the output of your system. For more information about expectations, check the section below.
If you're interested in the data the Scouts got through a measurement, you can tell Outpost that it must save that data after the measurement is run. That way you can inquiry it for further analysis. This way, you can have Scouts without any expectations/reports, so you can collect data without telling if the system is either up or down. You can check an usage example in the Reports integration test.
Expectations
Consider the following code snippet, taken from previous examples:
report :up, :response_code => 200
report :down, :response_body => {:match => /Ops/}
In the example above, :response_code and :response_body are expectations, responsible to get Scout's output and evaluate it, in order to determine a status.
They must be registered into each Scout that wish to support different types of expectations. You can supply a block or an object that respond to #call and return true if any of the rules match. It will receive an instance of the scout (so you can query current system state) as the first parameter and the state defined in the #report method as the second.
So you can easily create your own expectation. Let's recreate the :response_code in Outpost::Scouts::Http:
module Outpost
module Scouts
class Http < Outpost::Scout
expect(:response_code) { |scout,code| scout.response_code == code }
attr_reader :response_code
def setup(options)
@host = options[:host]
@port = options[:port] || 80
@path = options[:path] || '/'
end
def execute
response = Net::HTTP.get_response(@host, @path, @port)
@response_code = response.code.to_i
end
end
end
end
You can also check the supplied expectations in the source of the project to have an idea on how to implement more complex rules.
Notifiers
Notifiers query Outposts and act upon its status and reports. In the example below, an Email notifier is being used to report failures in the system to the system administrator:
require 'outpost'
require 'outpost/scouts'
require 'outpost/notifiers'
class HttpOutpostExample < Outpost::Application
notify Outpost::Notifiers::Email, {
:from => 'outpost@example.com',
:to => 'sleep_deprived_admin@example.com'
}
using Outpost::Scouts::Http => "web page" do
options :host => 'localhost', :port => 3000
report :up, :response_code => 200
report :down, :response_body => {:match => /Ops/}
end
end
outpost = HttpOutpostExample.new
outpost.run # => :down
# Will send an email to the poor sleep-deprived Sys Admin if the system is
# down.
outpost.notify if outpost.down?
Creating Outpost applications programatically
It is also possible to create Outposts without having to subclass it. Use the methods #add_scout and #add_notifier and you're set:
outpost = Outpost::Application.new
outpost.add_scout Outpost::Scouts::Http => 'master http server' do
options :host => 'localhost', :port => 9595
report :up, :response_code => 200
end
outpost.run
This is good when you want to have some sort of template (by inheriting from Outpost::Application) and then configure things as you go.
TODO
See TODO.
License
MIT License.