Forge Home

puppet_script

Imperative Puppet Scripts

6,432 downloads

6,432 latest version

4.1 quality score

We run a couple of automated
scans to help you access a
module's quality. Each module is
given a score based on how well
the author has formatted their
code and documentation and
modules are also checked for
malware using VirusTotal.

Please note, the information below
is for guidance only and neither of
these methods should be considered
an endorsement by Puppet.

Version information

  • 0.0.1 (latest)
released Apr 9th 2017
This version is compatible with:
  • Puppet Enterprise 2017.2.x, 2017.1.x, 2016.5.x, 2016.4.x
  • Puppet 4.x

Start using this module

  • r10k or Code Manager
  • Bolt
  • Manual installation
  • Direct download

Add this module to your Puppetfile:

mod 'binford2k-puppet_script', '0.0.1'
Learn more about managing modules with a Puppetfile

Add this module to your Bolt project:

bolt module add binford2k-puppet_script
Learn more about using this module with an existing project

Manually install this module globally with Puppet module tool:

puppet module install binford2k-puppet_script --version 0.0.1

Direct download is not typically how you would use a Puppet module to manage your infrastructure, but you may want to download the module in order to inspect the code.

Download

Documentation

binford2k/puppet_script — version 0.0.1 Apr 9th 2017

Puppet Script

Imperative scripts using the Puppet RAL

Puppet's state model is amazing when you're managing configuration. But sometimes you really just need to make a thing happen. Maybe you need to halt an application, update its database, then restart it. Maybe you need to take recovery actions when certain resources fail.

Puppet Script allows you to break out of the state model and just list resources states to manage in an imperative form. There's no dependency management, no duplicate resources to worry about, no immutable variables. Just write your script and let Puppet do its magic.

Reasons you might use this:

  • You need to orchestrate an application deployment or upgrade that involves stopping and restarting multiple services in the proper order.
  • Databases schema upgrades or data migrations need explicit orchestration.
  • You are transitioning from MySQL to PostgreSQL, or vice versa, and need to dump data, import into the new database and then dispose of the old database.
  • You need multiple levels of error handling, such as paging on-call support, initiating disaster recovery procedures, or failing over to a warm standby.
  • You need the run to fail immediately if any resources fail.

Disadvantages with this model:

  • This offers no consistency guarantee.

    When the script is done running, you know that each resource it enforced executed successfully in turn, but nothing more. For example, Puppet Script doesn't prevent you from managing a resource multiple times. You can see that in the example script below. As such, there is no single declaration of the complete state of any resource. Instead, it's some more-or-less indeterminate combination of the starting state and each step of the script.

  • There is very little visibility into changes.

    This means that if you manage a configuration file with a specific setting set there is nothing preventing you from accidentally managing it again with that setting unset. The last resource applied wins and there's no visibility anywhere else that this is happening.

  • The relationship model is sequential only.

    There are often relationships between different resources. For example, you cannot run a command until the package that contains that command has been installed. Puppet builds those relationships into it's core, making it very easy to ensure that all dependencies are met. In a script though, there's no way to indicate any of these relationships in code. Essentially, there's no way to know whether any of the resources that come afterwards depend on any given resource and there's no way for the system to validate or enforce the dependencies. It's left up to you, and the only way you can actually test it is to actually run the script.

  • There is no complete representation of the resulting configuration.

    A Puppet catalog is a complete representation of the entire configuration you care about. You could take that configuration and apply it on another representative system and come out with the same final result. You could inspect the catalog and know how it would configure a system. You could look at individual classes and resources in your codebase and have an expectation that they represent reality. That is not the case with a Puppet Script. There are no guardrails preventing you from managing a resource in multiple conflicting ways, and there's no way to read the code and have an expectation of what the result is without holding the entire codebase in your head.


To be clear: in almost every situation, you should not use this tool. This is only for complex processes which are difficult to represent as the final state of a state model or one-off ad hoc tasks. This should never be used for ongoing configuration management and it should never be used for scripts that are larger than you can hold in your head at once.

If you're considering using this because you're struggling with Puppet relationships, then please stop by https://slack.puppet.com/ or #puppet on Freenode. Someone there will be glad to help you solve your problem. Also refer to the documentation.


Writing a script

So what's it look like? It's just a Ruby based DSL. Write Ruby code with constructs like this to manage resources. Facts are available in the facts[] hash.

#! /opt/puppetlabs/bin/puppet script

['one', 'two', 'three'].each do |name|
  resource(:file, "/tmp/#{name}",
    :ensure  => 'file',
    :content => facts['osfamily'],
  )
end

resource(:file, '/tmp/dupe',
  :ensure  => 'file',
  :content => 'hello there',
)

resource(:file, '/tmp/dupe',
  :ensure => 'file',
  :mode   => '0600',
)

See the examples directory for more complex examples. For example, the examples/upgrade.pps script shows how a database-backed application could be upgraded, along with the database schema, with health checks and recovery if anything goes wrong.

The resource declaration works with any native types. It will not work for defined types or for including Puppet classes.

If you'd like to use defined resource types, or if you need to enforce some Puppet code for setup, then you can invoke the apply() method and directly enforce a mini-catalog of Puppet code. It's easiest to use a heredoc, as in this example:

#! /opt/puppetlabs/bin/puppet script

resource(:package, 'myapplication',
  :ensure  => present,
)

apply <<-EOS
include apache
apache::vhost { $facts['fqdn']:
  port    => '80',
  docroot => '/opt/myapplication/html',
}
notify { 'hello from puppet code': }
EOS

Running the script

Either make the script executable like any other script, or run it directly with puppet script. See puppet script --help for usage information.

root@master:~ # puppet script script.pps
File[/this/path/does/not/exist]
  - change from absent to file failed: Could not set 'file' on ensure: No such file or directory @ dir_s_mkdir - /this/path/does/not/exist20170316-7813-308f85.lock
Managed 7 resources with 1 failures.
root@master:~ # chmod +x script.pps
root@master:~ # ./script.pps
File[/this/path/does/not/exist]
  - change from absent to file failed: Could not set 'file' on ensure: No such file or directory @ dir_s_mkdir - /this/path/does/not/exist20170316-7813-308f85.lock
Managed 7 resources with 1 failures.

Installing

This is packaged as a Puppet module. Just drop it in your modulepath or put it in your Puppetfile. When I release it on the Forge, you can use puppet module install too.

Disclaimer

I take no liability for the use of this module. At this point, it's just a proof of concept.

Contact

binford2k@gmail.com