flapjack-admin

An admin interface for the Flapjack monitoring system.

You administer checks through flapjack-admin, then run flapjack-populator to place the checks on beanstalkd. A flapjack-worker will then pull the check off queue, and flapjack-notifier will notify if the check reports and bad result.

Dependencies (Ubuntu Hardy)

Install the following dependencies:

sudo apt-get install rubygems build-essential \
                     ruby-dev libxml2-dev \
                     libsqlite3-dev libxslt1-dev libopenssl-ruby

Setup

1. clone the repo
2. cd into cloned repo
3. bin/thor merb:gem:redeploy
4. MERB_ENV=production bin/rake db:automigrate (this will create a sqlite database for you to use)
5. bin/merb -e production
6. Go to http://localhost:4000/ in your browser

Now create some checks, and then see the next section about populating jobs.

Populating jobs

FIXME: this is deprecated, now the web interface handles this

Once you've created jobs in the admin interface, you want to pop them onto the worker queue. flapjack-admin provides a simple tool to do this for you.

To flush out the previous jobs, you need to restart the beanstalkd.

sudo /etc/init.d/beanstalkd stop
sudo /etc/init.d/beanstalkd start

Then, to populate jobs:

bin/flapjack-populator sqlite3://$(pwd)/production.db

Restarting beanstalkd from flapjack-admin

Once you've set up a bunch of checks in flapjack-admin, you'll want to push them onto your beanstalkd. flapjack-admin will attempt to do this for you when you create a batch of checks through the interface.

As there are many ways to deploy Flapjack (different distributions, security models, Passenger vs. Thin/Mongrel), it's up to the user to tell flapjack-admin how to restart the beanstalkd.

Don't fret, this is really easy. You need to specify the restart logic to Merb.config.restart_beanstalkd in config/environments/production.rb. There's example block in config/environments/development.rb, which is what I use to restart beanstalkd on my development machine.

Put simply, it looks something like this:

Merb::Config.use do |c|
  c[:restart_beanstalkd] = Proc.new do 
    # magic restart logic here!
  end
end

You can test this by running a rake task:

$ MERB_ENV='development' bin/rake test_beanstalkd_restart

Sub out development with test/production to test those environments.

This approach is a bit of a cop out, but you can't please all the people all the time. As flapjack-admin matures, a prefered way of restarting beanstalkd will emerge.

Testing

To run Cucumber features:

$ bin/cucumber features

The Cucumber features live under features/