Copyright © 2012-2013, Kannan Manickam
Client libraries for communicating with a Jenkins CI server and programatically managing jobs.
IRC Channel: ##jenkins-api-client (on freenode)
Mailing list: jenkins_api_client@googlegroups.com
Google Group: https://groups.google.com/group/jenkins_api_client
This project is a simple API client for interacting with Jenkins Continuous Integration server. Jenkins provides three kinds of remote access API.
- XML API, 2. JSON API, and 3. Python API. This project aims at consuming the JSON API and provides some useful functions for controlling jobs on the Jenkins programatically. Even though Jenkins provides an awesome UI for controlling jobs, it would be nice and helpful to have a programmable interface so we can dynamically and automatically manage jobs and other artifacts.
This projects currently only provides functionality for the jobs, node, view, system, and build queue interfaces.
Install jenkins_api_client by sudo gem install jenkins_api_client Include this gem in your code as a require statement.
require 'jenkins_api_client'
If you want to just play with it and not actually want to write a script, you can just use the irb launcher script which is available in scripts/login_with_irb.rb. But make sure that you have your credentials available in the correct location. By default the script assumes that you have your credentials file in ~/.jenkins_api_client/login.yml. If you don't prefer this location and would like to use a different location, just modify that script to point to the location where the credentials file exists.
ruby scripts/login_with_irb.rb
You will see the that it entered IRB session and you can play with the API client with the @client object that it has returned.
Supplying credentials to the client is optional, as not all Jenkins instances require authentication. This project supports two types of password-based authentication. You can just you the plain password by using password parameter. If you don't prefer leaving plain passwords in the credentials file, you can encode your password in base64 format and use password_base64 parameter to specify the password either in the arguments or in the credentials file. To use the client without credentials, just leave out the username and password parameters. The password parameter is only required if username is specified.
Support for Jenkins crumbs has been added. These allow an application to use the Jenkins API POST methods without requiring the 'Prevent Cross Site Request Forgery exploits' to be disabled. The API will check in with the Jenkins server to determine whether crumbs are enabled or not, and use them if appropriate.
As discussed earlier, you can either specify all the credentials and server information as parameters to the Client or have a credentials file and just parse the yaml file and pass it in. The following call just passes the information as parameters
@client = JenkinsApi::Client.new(:server_ip => '0.0.0.0',
:username => 'somename', :password => 'secret password')
# The following call will return all jobs matching 'Testjob'
puts @client.job.list("^Testjob")
The following example passes the YAML file contents. An example yaml file is located in config/login.yml.example.
@client = JenkinsApi::Client.new(YAML.Load_file(File.expand_path(
"~/.jenkins_api_client/login.yml", __FILE__)))
# The following call lists all jobs
puts @client.job.list_all
Sometimes we want certain jobs to be added as downstream projects and run them sequentially. The following example will explain how this could be done.
require 'jenkins_api_client'
# We want to filter all jobs that start with 'test_job'
# Just write a regex to match all the jobs that start with 'test_job'
jobs_to_filter = "^test_job.*"
# Create an instance to jenkins_api_client
@client = JenkinsApi::Client.new(YAML.load_file(File.expand_path(
"~/.jenkins_api_client/login.yml", __FILE__)))
# Get a filtered list of jobs from the server
jobs = @client.job.list(jobs_to_filter)
# Chain all the jobs with 'success' as the threshold
# The chain method will return the jobs that is in the head of the sequence
# This method will also remove any existing chaining
initial_jobs = @client.job.chain(jobs, 'success', ["all"])
# Now that we have the initial job(s) we can build them
# The build function returns a code from the API which should be 302 if
# the build was successful
code = @client.job.build(initial_jobs[0])
raise "Could not build the job specified" unless code == 302
In the above example, you might have noticed that the chain method returns an array instead of a single job. There is a reason behind it. In simple chain, such as the one in the example above, all jobs specified are chained one by one. But in some cases they might not be dependent on the previous jobs and we might want to run some jobs parallelly. We just have to specify that as a parameter.
For example: parallel = 3 in the parameter list to the chain method will take the first three jobs and chain them with the next three jobs and so forth till it reaches the end of the list.
There is another filter option you can specify for the method to take only jobs that are in a particular state. In case if we want to build only jobs that are failed or unstable, we can achieve that by passing in the states in the third parameter. In the example above, we wanted build all jobs. If we just want to build failed and unstable jobs, just pass ["failure", "unstable"]. Also if you pass in an empty array, it will assume that you want to consider all jobs and no filtering will be performed.
There is another parameter called threshold you can specify for the chaining and this is used to decide whether to move forward with the next job in the chain or not. A success will move to the next job only if the current build succeeds, failure will move to the next job even if the build fails, and unstable will move to the job even if the build is unstable.
The following call to the chain method will consider only failed and unstable jobs, chain then with 'failure' as the threshold, and also chain three jobs in parallel.
initial_jobs = @client.job.chain(jobs, 'failure', ["failure", "unstable"], 3)
# We will receive three jobs as a result and we can build them all
initial_jobs.each do |job|
code = @client.job.build(job)
raise "Unable to build job: #{job}" unless code == 302
end
To running Jenkins CLI
- authentication with username/password (deprecated)
@client = JenkinsApi::Client.new(:server_ip => '127.0.0.1',
:username => 'somename', :password => 'secret password')
# The following call will return the version of Jenkins instance
puts @client.exec_cli("version")
- authentication with public/private key file remember to upload the public key to http://:/user//configure
@client = JenkinsApi::Client.new(:server_ip => '127.0.0.1',
:identity_file => '~/.ssh/id_rsa')
# The following call will return the version of Jenkins instance
puts @client.exec_cli("version")
Before you run the CLI, please make sure the following requirements are fulfilled:
- JRE/JDK 6 (or above) is installed, and 'java' is on the $PATH environment variable
- The
jenkins_api_client/java_deps/jenkins-cli.jar
is required as the client to run the CLI. You can retrieve the available commands via accessing the URL:http://<server>:<port>/cli
- (Optional) required if you run the Groovy Script through CLI, make sure the user have the privilige to run script
Command line interface is supported only from version 0.2.0. See help using jenkinscli help
There are three ways for authentication using command line interface
- Passing all credentials and server information using command line parameters
- Passing the credentials file as the command line parameter
- Having the credentials file in the default location HOME/.jenkins_api_client/login.yml
As of v0.13.0, this debug parameter is removed. Use the logger instead. See the next section for more information about this option.
As of v0.13.0, support for logger is introduced. Since it would be nice to have the activities of the jenkins_api_client in a log file, this feature is implemented using the Ruby's standard Logger class. For using this feature, there are two new input arguments used during the initialization of Client.
:log_location
- This argument specifies the location for the log file. A good location for linux based systems would be '/var/log/jenkins_api_client.log'. The default for this values is STDOUT. This will print the log messages on the console itself.:log_level
- This argument specifies the level of messages to be logged. It should be one of Logger::DEBUG (0), Logger::INFO (1), Logger::WARN (2), Logger::ERROR (3), Logger::FATAL (4). It can be specified either using the constants available in the Logger class or using these integers provided here. The default for this argument is Logger::INFO (1)
If you want customization on the functionality Logger provides such as leave n old log files, open the log file in append mode, create your own logger and then set that in the client.
@client = JenkinsApi::Client.new(...)
# Create a logger which ages logfile once it reaches a certain size. Leave 10
# “old log files” and each file is about 1,024,000 bytes.
@client.logger = Logger.new('foo.log', 10, 1024000)
Please refer to Ruby Logger for more information.
If you would like to contribute to this project, just do the following:
- Fork the repo on Github.
- Add your features and make commits to your forked repo.
- Make a pull request to this repo.
- Review will be done and changes will be requested.
- Once changes are done or no changes are required, pull request will be merged.
- The next release will have your changes in it.
Please take a look at the issues page if you want to get started.
If you use this gem for your project and you think it would be nice to have a particular feature that is presently not implemented, I would love to hear that and consider working on it. Just open an issue in Github as a feature request.