Introduction

VagrantPy is a python module that provides a thin wrapper around the vagrant command line executable, allowing programmatic control of Vagrant virtual machines (boxes). This module is useful for:

• Starting a Vagrant virtual machine (VM) (up).
• Terminating a Vagrant VM (destroy).
• Halting a Vagrant VM without destroying it (halt).
• Querying the status of a VM or VMs (status).
• Getting ssh configuration information useful for SSHing into the VM. (host, port, ...)
• Running vagrant commands in a multi-VM environment (http://vagrantup.com/v1/docs/multivm.html) by using vm_name parameter.
• Initializing the VM based on a named base box, using init().
• Adding, Removing, and Listing boxes (box add, box remove, box list).
• Provisioning VMs - up() accepts options like no_provision, provision, and provision_with, and there is a provision() method.
• Using sandbox mode from the Sahara gem (https://github.com/jedi4ever/sahara).

This project began because I wanted python bindings for Vagrant so I could programmatically access my vagrant box using Fabric. Drop me a line to let me know how you use vagrantpy. I'd love to share more use cases. -Todd DeLuca

Versioning and API Stability

This package is beta and its API is not guaranteed to be stable. The API attempts to be congruent with the vagrant API terminology, to facilitate knowledge transfer for users already familiar with Vagrant. Over time, the vagrantpy API has changed to better match the underling vagrant CLI and to evolve with the changes in that CLI.

The package version numbering is in the form 0.X.Y. The initial 0 reflects the beta nature of this project. The number X is incremented when backwards-incompatible changes occur. The number Y is incremented when backwards-compatible features or bug fixes are added.

Requirements

• Vagrant 1.4 or greater (currently tested with 1.7.2). Using the latest version of Vagrant is strongly recommended.
• Vagrant requires VirtualBox, VMWare, or another supported provider.
• Python 2.7 (the only version this package has been tested with.) or Python 3.3 or higher.
• The Sahara gem for Vagrant is optional. It will allow you to use SandboxVagrant.

Installation

Install from pypi.python.org

Download and install vagrantpy:

pip install vagrantpy

Install from github.com

Clone and install vagrantpy

cd ~
git clone https://github.com/vagrantpy/vagrantpy.git
cd vagrantpy
python setup.py install

Usage

A contrived example of starting a vagrant box (using a Vagrantfile from the current directory) and running a fabric task on it:

import vagrant
from fabric.api import env, execute, task, run

@task
def mytask():
    run('echo $USER')


v = vagrant.Vagrant()
v.up()
env.hosts = [v.user_hostname_port()]
env.key_filename = v.keyfile()
env.disable_known_hosts = True # useful for when the vagrant box ip changes.
execute(mytask) # run a fabric task on the vagrant host.

Another example showing how to use vagrant multi-vm feature with fabric:

import vagrant
from fabric.api import *

@task
def start(machine_name):
   """Starts the specified machine using vagrant"""
   v = vagrant.Vagrant()
   v.up(vm_name=machine_name)
   with settings(host_string= v.user_hostname_port(vm_name=machine_name),
                 key_filename = v.keyfile(vm_name=machine_name),
                 disable_known_hosts = True):
        run("echo hello")

By default python vagrant instances are quiet, meaning that they capture stdout and stderr. For a "loud" instance, use vagrant.Vagrant(quiet_stdout=False). Set quiet_stderr=False for an even louder version.

Interacting With the Vagrant Subprocess

The Vagrant class works by executing vagrant commands in a subprocess and interpreting the output. Depending on the needs of the user, the communication to and from the subprocess can be tailored by altering its environment and where it sends its stdout and stderr.

Silencing the Stdout or Stderr of the Vagrant Subprocess

The stdout and stderr of the underlying vagrant process can be silenced by using the out_cm and err_cm parameters, or by using the quiet_stdout and quiet_stderr parameters of Vagrant.__init__.

Using out_cm and err_cm to redirect stdout and stderr to /dev/null:

v = vagrant.Vagrant(out_cm=vagrant.devnull_cm, err_cm=vagrant.devnull_cm)
v.up() # normally noisy

Using quiet_stdout and quiet_stderr to redirect stdout and stderr to /dev/null:

v = vagrant.Vagrant(quiet_stdout=True, quiet_stderr=True)
v.up() # normally noisy

These are functionally equivalent.

Logging the Stdout or Stderr of the Vagrant Subprocess

A user might wish to direct the stdout and stderr of a vagrant subprocess to a file, perhaps to log and analyze the results of an automated process. This can be accomplished using the out_cm and err_cm parameters of Vagrant.__init__.

For example, log the stdout and stderr of the subprocess to the file 'deployment.log':

log_cm = vagrant.make_file_cm('deployment.log')
v = vagrant.Vagrant(out_cm=log_cm, err_cm=log_cm)
v.up() # normally noisy

Altering the Environment of the Vagrant Subprocess

It's possible to communicate with the Vagrant subprocess using environment variables. The Vagrantfile could expect environment variables to be present and act accordingly. The environment variables can be set by vagrantpy.

import vagrant

v = vagrant.Vagrant()

os_env = os.environ.copy()
os_env['USE_NFS'] = '1'

v.env = os_env
v.up()  # will pass env to the vagrant subprocess

Alternatively, the environment can be passed at instantiation time.

import vagrant

os_env = os.environ.copy()
os_env['USE_NFS'] = '1'

v = vagrant.Vagrant(env=env)
assert v.env is env  # True
v.up()  # will pass env to the vagrant subprocess

Contribute

If you use python and vagrant and this project does not do what you want, please open an issue or a pull request on github at https://github.com/vagrantpy/vagrantpy.

Please see CHANGELOG.md for a detailed list of contributions and authors.

When making a pull request, please include unit tests that test your changes and make sure any existing tests still work. See the Testing section below.

Testing

Running the full suite of tests might take 10 minutes or so. It involves downloading boxes and starting and stopping virtual machines several times.

Run the tests from the top-level directory of the repository:

nosetests

Here is an example of running an individual test:

nosetests tests.test_vagrant:test_boxes

Manual test of functionality for controlling where the vagrant subcommand output is sent -- console or devnull:

>>> import vagrant
>>> import os
>>> vagrantfile = '/Users/tfd/proj/vagrantpy/tests/vagrantfiles/single_box'
>>> # Demonstrate a quiet Vagrant.  Equivalent to out_cm=vagrant.devnull_cm
... v1 = vagrant.Vagrant(vagrantfile)
>>> v1.destroy() # output to /dev/null
>>> # Demonstrate a loud Vagrant.  Equivalent to out_cm=vagrant.stdout_cm
... v2 = vagrant.Vagrant(vagrantfile, quiet_stdout=False)
>>> v2.destroy() # stdout sent to console
==> default: VM not created. Moving on...
>>> # Demonstrate that out_cm takes precedence over quiet_stdout=True
... v3 = vagrant.Vagrant(vagrantfile, out_cm=vagrant.stdout_cm)
>>> v3.destroy() # output to console
==> default: VM not created. Moving on...
>>> # Demonstrate a quiet Vagrant using devnull_cm directly
... v4 = vagrant.Vagrant(vagrantfile, out_cm=vagrant.devnull_cm)
>>> v4.destroy() # output to console
>>>