Tiny Puppet

The Universal Installer

Tiny Puppet is single Puppet module that manages virtually any application on any Operating System:

puppet module install example42-tp
puppet tp setup
tp install <any_app>

It can be used inside Puppet manifests or directly from the command line.

Features:

• Quick, easy to use, standard, coherent, powerful interface to applications installation and their config files management.

• Out of the box and easily expandable support for most common Operating Systems

• Modular data source design. Support for an easily growing set of applications.

• Smooth coexistence with any existing Puppet modules setup: you decide what to manage

• Application data stored in a configurable separated module (tinydata is the default source for applications data)

• A Puppet face and command line to install any app or query the installed ones

It is intended to be used in profiles, as replacement for dedicated componenent modules, or in the same modules, to ease the management of the provided files and packages.

The expected users are both experienced sysadmins who know exactly how to configure their applications and absolute beginners who want to simply install an application, without knowing how it's package is called on the underlying system or how to install its repositories or dependencies.

Provided Puppet defines

Tiny Puppet provides the following Puppet user defines:

• tp::install. It installs an application and starts its service, by default
• tp::conf. It allows to manage configuration files
• tp::dir. Manages the content of directories
• tp::stdmod. Manages the installation of an application using StdMod compliant parameters
• tp::test. Allows quick and easy (acceptance) testing of an application
• tp::repo. Manages extra repositories for the supported applications
• tp::puppi. Puppi integration (Don't worry, fully optional)
• tp::concat. (WIP) Manages file fragments of a configuration file
• tp::netinstall. (WIP) Installs from a remote url
• tp::instance. (TODO) Manages an application instance
• tp::github. (TODO?) Installs (anything?;) directly from GitHub source

Usage on the command line

Tiny Puppet adds the tp command to Puppet. Just have it in your modulepath and install the tp command with:

puppet tp setup

With the tp command you can install on the local OS the given application, taking care of naming differences, additional repos or prerequisites.

tp install <application>
tp uninstall <application>

tp test # Test all the applications installed by tp
tp test <application> # Test the specified application

tp log # Tail all the logs of all the applications installed by tp
tp log <application> # Tail the log of the specified application

Each of these commands can be inkoed also via the tp puppet face:

puppet tp <command> <arguments>

Prerequisites and limitations

Main Tiny Puppet defines work only on Puppet 4.x but the module provides also Puppet 3.x compatible defines with the 3 suffix.

So, for example:

tp::install   # Is optimised for Puppet 4 (doesn't work on earlier versions)
tp::install3  # Works on Puppet 3 and 4

IMPORTANT NOTE: Do not expect all the applications to flawlessly work out of the box for all the Operating Systems. Tiny Puppet manages applications that can be installed and configured using the underlying OS native packages and services, this might not be possible for all the cases.

If tp doesn't correctly install a specific application on the OS you want, please TELL US. It's very easy and quick to add new apps or support for new OS, we will do that.

Currently most of the applications are supported on RedHat and Debian derivatives Linux distributions.

Support for Solaris, Windows, MacOS, *BSD and others is currently limited, mostly for lack of tinydata.

Tiny Puppet requires these Puppet modules:

• The tinydata module

• Puppet Labs' stdlib module.

If you use the relevant defines, other dependencies are needed:

• Define tp::concat requires Puppet Labs' concat module.

• Define tp::dir , when used with the vcsrepo argument, requires Puppet Labs' vcsrepo module.

• Define tp::puppi requires Example42's puppi module.

Usage in Puppet code

Install an application with default settings (package installed, service started)

tp::install { 'redis': }

Configure the application main configuration file a custom erb template which uses data from a custom $options_hash:

tp::conf { 'rsyslog':
  template     => 'site/rsyslog/rsyslog.conf.erb',
  options_hash => hiera('rsyslog::options_hash'),
}

Populate any custom directory from a Git repository (it requires Puppet Labs' vcsrepo module):

tp::dir { '/opt/apps/my_app': 
  source      => 'https://git.example.42/apps/my_app/',
  vcsrepo     => 'git',
}

Installation options

Install custom packages (with the settings_hash argument you can override any application specific setting)

tp::install { 'redis':
  settings_hash => {
    'package_name'     => 'my_redis',
    'config_file_path' => '/opt/etc/redis',
  },
}

Some options are available to manage tp::install automation:

tp::install { 'virtualbox':
  auto_repo          => true,  # This is the default, settings, if defined in tinydata, it installs the relevant package repository
  auto_conf          => true,  # True by default. If defined in tinydata a default configuration is provided
  auto_prerequisites => false, # False by default. If true eventual package or tp::install dependencies are installed
}

Other options are available to manage integrations:

tp::install { 'rabbitmq':
  cli_enable    => true,  # Default value. Installs the tp command on the system and provides the data about the defined application.
  puppi_enable  => false, # Default value. Installs puppi and enables puppet integration
  test_template => undef, # Default value. If provided, the provided erb template is used as script to test the application (instead of default tests)
  options_hash  => {}     # An optional hash where to set variable to use in the test_template.
}

To uninstall an application, there are two alternatives:

tp::uninstall { 'redis': }
tp::install { 'redis': ensure => absent }

Installation alternatives

To manage packages installations and configuration files there's also the tp::stdmod define to manage an application using stdmod compliant parameters.

Note that tp::stdmod is alternative to tp::install (both of them manage packages and services) and may be complementary to tp::conf (you can configure files with both).

tp::stdmod { 'redis':
  config_file_template => 'site/redis/redis.conf',
}

If you wonder what's better, use tp::install + tp::conf rather than tp::stdmod.

Managing configurations

By default, configuration files managed by tp::conf automatically notify the service(s) and require the package(s) installed via tp::install. If you use tp::conf without a relevant tp::install define and have dependency cycle problems or references to non existing resources, you can disable these automatic relationships:

tp::conf { 'bind':
  config_file_notify  => false,
  config_file_require => false,
}

You can also set custom resource references to point to actual resources you declare in your manifests:

tp::conf { 'bind':
  config_file_notify  => Service['bind9'],
  config_file_require => Package['bind9-server'],
}

It's possible to manage files with different methods, for example directly providing its content:

tp::conf { 'redis':
  content => 'my content is king',
}

or providing a custom erb template (used as content => template($template)):

tp::conf { 'openssh::ssh_config':
  template    => 'site/openssh/ssh_config.erb',
}

or using a custom epp template with Puppet code instead of Ruby (used as content => epp($epp)):

tp::conf { 'redis:
  epp   => 'site/redis/redis.conf.epp',
}

also it's possible to provide the source to use, instead of managing it with the content argument:

tp::conf { 'redis':
  source      => [ "puppet:///modules/site/redis/redis.conf-${hostname}" ,
                   'puppet:///modules/site/redis/redis.conf' ] ,
}

Tp:conf has some conventions on the actual configuration file manages.

By default, if you just specify the application name, the file managed is the "main" configuration file of that application (in case this is not evident or may be questionable, check the data files for the actual value used).

# This manages /etc/ssh/sshd_config
tp::conf { 'openssh':
  [...]
}

If you specify a file name after the application name in the title, separated by ::, and you don't specify any alternative base_file, then that file is placed in the "base" configuration dir:

# This manages /etc/ssh/ssh_config
tp::conf { 'openssh::ssh_config':
  [...]
}

If you specify the parameter base_file then the path is the one of the specified base_file and the title does not provide any information about the managed file path (it still needs the relevant app in the first part, before ::, and it needs to be unique across the catalog).

# This manages /etc/default/puppetserver on Debian or /etc/sysconfig/puppetserver on RedHat
tp::conf { 'puppetserver::init':
  base_file => 'init',
  [...]
}

If you explicitly set a path, that path is used and the title is ignored (be sure, anyway, to refer to a supported application and is not duplicated in your catalog):

# This manages /usr/local/bin/openssh_check
tp::conf { 'openssh::ssh_check':
  path => '/usr/local/bin/openssh_check',
  [...]
}

Managing directories

Manage a whole configuration directory:

tp::dir { 'redis':
  source      => 'puppet:///modules/site/redis/',
}

Manage a specific directory type. Currently defined directories types are:

• config : The application [main] configuration directory (Default value)
• conf : A directory where you can place single configuration files (typically called ./conf.d )
• data : Directory where application data resides
• log : Dedicated directory for logs, if present

Note that some of these directory types might not be defined for every application.

tp::dir { 'apache':
  base_dir => 'data',
  source   => 'puppet:///modules/site/apache/default_site',
}

Clone a whole configuration directory from a Git repository (it requires Puppet Labs' vcsrepo module):

tp::dir { 'redis':
  source      => 'https://git.example.42/puppet/redis/conf/',
  vcsrepo     => 'git',
}

Populate any custom directory from a Subversion repository (it requires Puppet Labs' vcsrepo module):

tp::dir { 'my_app': # The title is irrilevant, when path argument is used 
  path        => '/opt/apps/my_app',
  source      => 'https://svn.example.42/apps/my_app/',
  vcsrepo     => 'svn',
}

Provide a data directory (the default DocumentRoot, for apache) from a Git repository (it requires Puppet Labs' vcsrepo module) (TODO):

tp::dir { 'apache':
  # base_dir is a tag that defines the type of directory for the specified application.
  # Default: config. Other possible dir types: 'data', 'log', 'confd', 'lib'
  # or any other name defined in the application data with a format like: ${base_dir}_dir_path
  base_dir    => 'data' 
  source      => 'https://git.example.42/apps/my_app/',
  vcsrepo     => 'git',
}

Managing repositories

Currently Tiny Puppet supports applications' installation only via the OS native packaging system. In order to cope with software which may not be provided by default on an OS, TP provides the tp::repo define that manages YUM and APT repositories for RedHat and Debian based Linux distributions.

The data about a repository is managed as all the other data of Tiny Puppet. Find here an example for managing Apt repositories and here one for Yum ones.

Generally you don't have to use directly the tp::repo define, as, when the repository data is present, it's automatically added from the tp::install one.

When it's present the relevant data for an application, it's possible to install it using different alternative repos. For example you can use:

tp::install { 'mongodb':
  repo => 'mongodb-org-3.2',
}

to install MongoDB using packages from the 3.2 upstream repo, instead of the default OS ones.

In some cases, where for the given application name there are no packages, the following commands have exactly the same effect:

tp::install { 'epel': }  # Installs Epel repository on RedHat derivatives. Does nothing on other OS.
tp::repo { 'epel': }     # Same effect of tp::install since no package is actually installed

If, for whatever reason, you don't want to automatically manage a repository for an application, you can set to false the auto_repo parameter, and, eventually you can manage the repository in a custom dependency class:

tp::install { 'elasticsearch':
  auto_repo        => false,
}

Using alternative data sources

By default Tiny Puppet uses the tinydata module to retrieve data for different applications, but it's possible to use a custom one:

tp::install { 'apache':
  data_module => 'my_data', # Default: tinydata
}

Reproduce the structure of tinydata in your data module to make it work with tp.

If you want to use your own data module for all your applications, you might prefer to set the following resource defaults in your main manifest (manifest/site.pp, typically):

Tp::Install {
  data_module  => 'my_data',
}
Tp::Conf {
  data_module  => 'my_data',
}
Tp::Dir {
  data_module  => 'my_data',
}

Testing and playing with Tiny Puppet

You can experiment and play with Tiny Puppet and see a lot of use examples on Example42 Puppet control-repo.

Acceptance tests are regularly done to verify tp support for different applications on different Operating Systems. They are in the TP acceptance repo.

Check this Compatibility Matrix for a quick overview on how different applications are currently supported on different Operating Systems.