Puppet

Puppet

• is a configuration management tool
• has it's own language
• ships with multiple utilities (reports, logs, web interfaces, etc...)
• Core is Open Source
• Some utilities aren't (and are included in the payed version)
• Works for Linux, Unix, even windows to some degree
• Is used by Sony, RedHat, Intel, Github, Cisco...

Basic idea of configuration management tools:

• Infrastructure as Code: Track, Test, Deploy, Reproduce, Scale
• Reproducible setups: Do once, repeat forever
• Scale quickly: Done for one, use on many
• Coherent and consistent server setups
• Continuous integration is usually implied

Puppet setup (basic)

• Create a puppet manifest (aka configuration file)
• Get it to the server somehow (Git, rsync, usb)
• Run the configuration (assuming you have puppet installed on the server)

Puppet setup (master - slave)

• A "master" containing the configuration files
• Clients which ask the master periodically (cron) or wait for updates (daemonization)
• All config files are @ one place

A minimal configuration file

user { 'zlatin':
  ensure     => present,
  uid        => '1000',
  gid        => '1000',
  shell      => '/bin/zsh',
  home       => '/home/zlatin'
}

There's a lot of gritty details to setting up puppet!

Example CI setup

Git

puppet master

Node

Node

On every push request:

• run tests /compile (fail if they don't pass)

After push request:

• update puppet master repo
• send update to daemons

Questions?