Rack provides a minimal, modular and adaptable interface
for developing web applications in Ruby.
By wrapping HTTP requests and responses in the simplest way possible,
it unifies and distills the API for web servers, web frameworks, and
software in between (the so-called middleware) into a single method call.
- app-servers ›
Rack provides a minimal, modular and adaptable interface for developing web applications in Ruby. This Charm will deploy Ruby on Rails, Sinatra or any other Rack application and connect it to supported services.
To deploy this charm you will need at a minimum: a cloud environment, working Juju installation and a successful bootstrap. Once bootstrapped, deploy Rack charm and all required services.
Ruby on Rails example
Configure your application, for example:
sample-rails: repo: https://github.com/pavelpachkovskij/sample-rails
Deploy Rack application:
juju deploy rack sample-rails --config sample-rails.yml
Deploy and relate database
juju deploy postgresql juju add-relation postgresql:db sample-rails
Now you can run migrations:
juju ssh sample-rails/0 run rake db:migrate
juju ssh sample-rails/0 run rake db:seed
And finally expose the Rack service:
juju expose rack
Find the Rack instance's public URL from
juju deploy mysql juju add-relation mysql rack
Configure your application, for example html2haml
html2haml: repo: https://github.com/twilson63/html2haml.git
Deploy your Rack service
juju deploy rack html2haml --config html2haml.yml
Expose Rack service:
juju expose html2haml
Source code updates
juju set <service_name> revision=<revision>
juju ssh <unit_name> run <command>
juju ssh <unit_name> sudo restart rack
You can add Procfile to your application and Rack to start additional processes or replace default application server:
web: bundle exec unicorn -p $PORT watcher: bundle exec rake watch
Specifying a Ruby Version
You can use the ruby keyword of your app's Gemfile to specify a particular version of Ruby.
source "https://rubygems.org" ruby "1.9.3"
Juju makes it easy to scale your Rack application. You can simply deploy any supported load balancer, add relation and launch any number of application instances.
juju deploy rack rack --config rack.yml juju deploy haproxy juju add-relation haproxy rack juju expose haproxy juju add-unit rack -n 2
Apache2 is harder to start with, but it provides more flexibility with configuration options.
Here is a quick example of using Apache2 as a load balancer with your rack application:
Deploy Rack application
juju deploy rack --config rack.yml
You have to enable mod_proxy_balancer and mod_proxy_http modules in your Apache2 config:
apache2: enable_modules: proxy_balancer proxy_http
juju deploy apache2 --config apache2.yml
Create balancer relation between Apache2 and Rack application
juju add-relation apache2:balancer rack
Apache2 charm expects a template to be passed in. Example of vhost that will balance all traffic over your application instances:
<VirtualHost *:80> ServerName rack ProxyPass / balancer://rack/ lbmethod=byrequests stickysession=BALANCEID ProxyPassReverse / balancer://rack/ </VirtualHost>
Update Apache2 service config with this template
juju set apache2 "vhost_http_template=$(base64 < vhost.tmpl)"
Expose Apache2 service
juju expose apache2
Logging with Logstash
You can add logstash service to collect information from application's logs and Kibana application to visualize this data.
juju deploy kibana juju deploy logstash-indexer juju add-relation kibana logstash-indexer:rest juju deploy logstash-agent juju add-relation logstash-agent logstash-indexer juju add-relation logstash-agent rack juju set logstash-agent CustomLogFile="['/var/www/rack/current/log/*.log']" CustomLogType="rack" juju expose kibana
Monitoring with Nagios and NRPE
You can can perform HTTP checks with Nagios. To do this deploy Nagios and relate it to your Rack application:
juju deploy nagios juju add-relation rack nagios
Additionally you can perform disk, mem, and swap checks with NRPE extension:
juju deploy nrpe juju add-relation rack nrpe juju add-relation nrpe nagios
Deploy MonogDB service and relate it to Rack application:
juju deploy mongodb juju add-relation mongodb rack
Rack charm will set environment variables which you can use to configure your Mongodb adapter.
MONGODB_URL => mongodb://host:port/database
Your mongoid.yml should look like:
production: uri: <%= ENV['MONGODB_URL'] %>
Your mongoid.yml should look like:
production: sessions: default: uri: <%= ENV['MONGODB_URL'] %>
In both cases you can set additional options specified by Mongoid.
Deploy Memcached service and relate it to Rack application:
juju deploy memcached juju add-relation memcached rack
Rack charm will set environment variables which you can use to configure your Memcache adapter. Dalli use those variables by default.
MEMCACHE_PASSWORD => xxxxxxxxxxxx MEMCACHE_SERVERS => instance.hostname.net MEMCACHE_USERNAME => xxxxxxxxxxxx
Deploy Redis service and relate it to Rack application:
juju deploy redis-master juju add-relation redis-master:redis-master rack
Rack charm will set environment variables which you can use to configure your Redis adapter.
REDIS_URL => redis://username:email@example.com:6389
For example you can configure Redis adapter in config/initializers/redis.rb
uri = URI.parse(ENV["REDIS_URL"]) REDIS = Redis.new(:host => uri.host, :port => uri.port, :password => uri.password)
Rack application didn't start because assets were not compiled
To be able to compile assets before you've joined database relation you have to disable initialize_on_precompile option in application.rb:
config.assets.initialize_on_precompile = false
If you can't do this you still can join database and compile assets manually:
juju ssh rack/0 run rake assets:precompile
Then restart Rack service (while you have to replace 'rack/0' with your application name, e.g. 'sample-rails/0', 'sudo restart rack' is a valid command to restart any deployed application):
juju ssh rack/0 sudo restart rack
Deploy from Git
Sample Git config:
rack: repo: <repository_url> revision: <revision_number>
To deploy from private repo via SSH add 'deploy_key' option:
Deploy from SVN
Sample SVN config:
rack: scm_provider: svn repo: <repository_url> revision: <revision_number> svn_username: <username> svn_password: <password>
Install extra packages
Specify list of packages separated by spaces:
extra_packages: 'libsqlite3++-dev libmagick++-dev'
Set ENV variables
You can set ENV variables, which will be available within all processes defined in a Procfile:
env: 'AWS_ACCESS_KEY_ID=aws_access_key_id AWS_SECRET_ACCESS_KEY=aws_secret_access_key'
- (string) A deploy key is an SSH key that is stored on the server and grants access to a repository (git only).
- (string) List of environment variables (e.g. SECRET_TOKEN=secret_token). These variables will override those which were set by Rack charm (like DATABASE_URL, REDIS_URL)
- (string) Extra packages to install before bundle install
- (string) Both RACK_ENV and RAILS_ENV environment variables.
- (string) Application repository URL
- (string) The revision to be checked out. This can be symbolic, like HEAD or it can be a source control management-specific revision identifier.
- (string) The name of the source control management provider to be used (git or svn).
- (string) The user name for a user that has access to the Subversion repository (svn only).
- (string) The password for the user that has access to the Subversion repository (svn only).