This is Janky, a continuous integration server built on top of Jenkins, controlled by Hubot, and designed for GitHub.
-
Built on top of Jenkins. The power, vast amount of plugins and large community of the popular CI server all wrapped up in a great experience.
-
Controlled by Hubot. Day to day operations are exposed as simple Hubot commands that the whole team can use.
-
Designed for GitHub. Janky creates the appropriate web hooks for you and the web app restricts access to members of your GitHub organization.
Start by setting up a new Jenkins job and GitHub web hook for a repository:
hubot ci setup github/janky
The setup
command can safely be run over and over again. It won't do
anything unless it needs to. It takes an optional name argument:
hubot ci setup github/janky janky-ruby1.9.2
All branches are built automatically on push. Disable auto build with:
hubot ci toggle janky
Run the command again to re-enable it. Force a build of the master branch:
hubot ci build janky
Of a specific branch:
hubot ci build janky/libgit2
Different builds aren't relevant to the same Campfire room and so Janky lets you choose where notifications are sent to. First get a list of available rooms:
hubot ci rooms
Then pick one:
hubot ci set room janky The Serious Room
Get the status of a build:
hubot ci status janky
Specific branch:
hubot ci status janky/libgit2
All builds:
hubot ci status
Finally, get a quick reference of the available commands with:
hubot ci?
Janky requires access to a Jenkins server. Version 1.427 is recommended. Refer to the Jenkins documentation for installation instructions and install the Notification Plugin version 1.4.
Remember to set the Jenkins URL in http://your-jenkins-server.com/configure
.
Janky will still trigger builds but will not update the build status without this set.
Janky is designed to be deployed to Heroku.
Grab all the necessary files from the gist:
$ git clone git://gist.github.com/1497335 janky
Then push it up to a new Heroku app:
$ cd janky
$ heroku create --stack cedar
$ bundle install
$ git add Gemfile.lock
$ git commit Gemfile.lock -m "lock bundle"
$ git push heroku master
After configuring the app (see below), create the database:
$ heroku run rake db:migrate
NOTE: Ruby version 1.9.3 is required to run Janky.
We strongly recommend backing up your Janky database before upgrading.
The general process is to then upgrade the gem, and then run migrate. Here is how you do that on a local box you have access to (this process will differ for Heroku):
cd [PATH-TO-JANKY]
gem update janky
rake db:migrate
Janky is configured using environment variables. Use the heroku config
command:
$ heroku config:add VARIABLE=value
Required settings:
JANKY_BASE_URL
: The application URL with a trailing slash. Example:http://mf-doom-42.herokuapp.com/
.JANKY_BUILDER_DEFAULT
: The Jenkins server URL with a trailing slash. Example:http://jenkins.example.com/
. For basic auth, include the credentials in the URL:http://user:[email protected]/
. Using GitHub OAuth with Jenkins is not supported by Janky.JANKY_CONFIG_DIR
: Directory where build config templates are stored. Typically set to/app/config
on Heroku.JANKY_HUBOT_USER
: Login used to protect the Hubot API.JANKY_HUBOT_PASSWORD
: Password for the Hubot API.JANKY_GITHUB_USER
: The login of the GitHub user used to access the API. Requires Administrative privileges to set up service hooks.JANKY_GITHUB_PASSWORD
: The password for the GitHub user.JANKY_GITHUB_HOOK_SECRET
: Secret used to sign hook requests from GitHub.JANKY_CHAT_DEFAULT_ROOM
: Chat room where notifications are sent by default.
Optional database settings:
DATABASE_URL
: Database connection URL. Example:postgres://user:password@host:port/db_name
.JANKY_DATABASE_SOCKET
: Path to the database socket. Example:/var/run/mysql5/mysqld.sock
.
Using Janky with GitHub Enterprise requires one extra setting:
JANKY_GITHUB_API_URL
: Full API URL of the instance, with a trailing slash. Example:https://github.example.com/api/v3/
.
https://github.com/blog/1227-commit-status-api
To update pull requests with the build status generate an OAuth token via the GitHub API:
curl -u username:password \
-d '{ "scopes": [ "repo:status" ], "note": "janky" }' \
https://api.github.com/authorizations
then set JANKY_GITHUB_STATUS_TOKEN
.
username
and password
in the above example should be the same as the
values provided for JANKY_GITHUB_USER
and JANKY_GITHUB_PASSWORD
respectively.
Janky notifies Campfire chat rooms by default. Required settings:
JANKY_CHAT_CAMPFIRE_ACCOUNT
: account name.JANKY_CHAT_CAMPFIRE_TOKEN
: authentication token for the user sending build notifications.
Required settings:
JANKY_CHAT=hipchat
JANKY_CHAT_HIPCHAT_TOKEN
: authentication token (This token needs to be an admin token, not a notification token.)JANKY_CHAT_HIPCHAT_FROM
: name that messages will appear be sent from. Defaults toCI
.JANKY_HUBOT_USER
should be XMPP/Jabber username in format xxxxx_xxxxxx rather than emailJANKY_CHAT_DEFAULT_ROOM
should be the name of the room instead of the XMPP format, for example:Engineers
instead of xxxx_xxxxxx.
Installation:
- Add
require "janky/chat_service/hipchat"
to theconfig/environment.rb
file before theJanky.setup(ENV)
line. echo 'gem "hipchat", "~>0.4"' >> Gemfile
bundle
git commit -am "install hipchat"
To restrict access to members of a GitHub organization, register a new
OAuth application on GitHub
with the callback set to $JANKY_BASE_URL/auth/github/callback
then set
a few extra settings:
JANKY_SESSION_SECRET
: Random session cookie secret. Typically generated by a tool likepwgen
.JANKY_AUTH_CLIENT_ID
: The client ID of the OAuth application.JANKY_AUTH_CLIENT_SECRET
: The client secret of the OAuth application.JANKY_AUTH_ORGANIZATION
: The organization name. Example: "github".JANKY_AUTH_TEAM_ID
: An optional team ID to give auth to. Example: "1234".
Install the janky script in your Hubot
then set the HUBOT_JANKY_URL
environment variable. Example:
http://user:[email protected]/_hubot/
, with user and password
replaced by JANKY_HUBOT_USER
and JANKY_HUBOT_PASSWORD
respectively.
The default build command should suffice for most Ruby applications:
$ bundle install --path vendor/gems --binstubs
$ bundle exec rake
For more control you can add a script/cibuild
at the root of your
repository for Jenkins to execute instead.
For total control, whole Jenkins' config.xml
files can be associated
with Janky builds. Given a build called windows
, Janky will try
config/jobs/windows.xml.erb
before falling back to the default
configuration, config/jobs/default.xml.erb
. After updating or adding
a custom config, run hubot ci setup
again to update the Jenkins
server.
Get your environment up and running:
script/bootstrap
Create the databases:
mysqladmin -uroot create janky_development
mysqladmin -uroot create janky_test
Create the tables:
RACK_ENV=development bin/rake db:migrate
RACK_ENV=test bin/rake db:migrate
Seed some data into the development database:
bin/rake db:seed
Start the server:
script/server
Open the app:
open http://localhost:9393/
Run the test suite:
script/test
Fork the Janky repository on GitHub and send a Pull Request. Note that any changes to behavior without tests will be rejected. If you are adding significant new features, please add both tests and documentation.
Copyright © 2011-2013, GitHub, Inc. See the COPYING
file for license
rights and limitations (MIT).