Skip to content

spark-jobserver/chef-spark-jobserver

Repository files navigation

spark-jobserver cookbook

Installs the Spark Jobserver.

Dev quickstart:

# minimal required configuration; these values have no defaults and must be provided
chef.json = {
  spark: {
    jobserver: {
      # path to your spark install
      spark_home: '/usr/local/spark',
      spark_conf_dir: '/usr/local/spark/conf',
      # you need to provide the jobserver jar; this gets downloaded, but can be a `file://` uri
      jar_url: 'https://domain.com/some/path/spark-job-server.jar'
    }
  }
}

include_recipe 'spark-jobserver::default'

To use the provided Vagrantfile to spin up a demo instance:

  1. Copy the Spark distribution binary into the repo root. By default this is assumed to be spark-1.1.0-bin-hadoop2.4.tgz. See below if you wish to use a different Spark distro.
  2. Copy the jobserver jar into the repo root. By default this is assumed to be spark-job-server.jar. Building the jobserver itself is easy: pushd .../spark-jobserver; sbt clean assembly.
  3. Spin up the Vagrant instance: vagrant up
  • If you want to use a different Spark distro, set the SPARK_DIST environment variable (without the extension): SPARK_DIST=spark-1.1.0-bin-cdh4 vagrant up
  1. Now you can browse the Jobserver UI at http://33.33.33.123:8090 and go through the jobserver examples.

Note that this repo assumes RVM; YMMV with other installations such as ChefDK. The Vagrant demo requires following plugins:

Supported Platforms

Chefspec tested on:

{
  'ubuntu' => [ '12.04', '13.04', '13.10', '14.04' ],
  'debian' => [ '6.0.5' ],
  'centos' => [ '5.8', '6.0', '6.3' ],
  'redhat' => [ '5.8', '6.3' ],
}

Confirmed to work on:

  • Ubuntu 12.04
  • Ubuntu 14.04

With Chef 11.8+

Attributes

See attributes for all options.

Usage

The only cookbook dependency is runit; java is assumed to be installed somehow.

Follow the appropriate guides for Spark standalone or Mesos deployment. The relevant attributes are:

  • node.spark.jobserver.master_type
  • node.spark.jobserver.master_url

spark-jobserver::default

Include spark-jobserver::default in your node's run_list:

{
  "run_list": [
    "recipe[spark-jobserver::default]"
  ]
}

Contributing

  1. Fork the repository on Github
  2. Create a named feature branch (i.e. add-new-recipe)
  3. Write your change
  4. Write tests for your change (if applicable)
  • At least a Chefspec proving the fix; Test Kitchen coming later
  1. Run the tests, ensuring they all pass
  2. Submit a Pull Request

License and Authors

Author:: Matt Chu ([email protected])

About

Chef recipe for Spark JobServer

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages