terraform-provider-vagrant
terraform-provider-vagrant copied to clipboard
A Vagrant provider for terraform.
terraform-provider-vagrant
A Vagrant provider for terraform 0.12+.
A note about lippertmarkus/vagrant in the registry: when I originally wrote this provider, the terraform registry didn't exist. My terraform needs waned and I didn't hear about the registry until some time later. lippertmarkus forked my provider and published to the registry as a convenience. Thanks! But, it's just an older version of this exact same codebase. So, I recommend you use bmatcuk/vagrant to get the latest updates instead.
Installation
Add bmatcuk/vagrant to required_providers:
terraform {
required_providers {
vagrant = {
source = "bmatcuk/vagrant"
version = "~> 4.0.0"
}
}
}
Usage
resource "vagrant_vm" "my_vagrant_vm" {
name = "vagrantbox"
vagrantfile_dir = "path/to/dir"
env = {
KEY = "value",
}
get_ports = true
}
name (optional) if the name changes, it will force the resource to destroy and recreate. Defaults to "vagrantbox".
vagrantfile_dir is the path to a directory where a Vagrantfile lives. The
Vagrantfile must exist when terraform runs or else it will throw an error. This
option defaults to .
, ie, the current directory and you may set this value to
absolute or relative paths.
env (optional) is a map of additional environment variables to pass to the Vagrantfile. The environment variables set by the calling process are always passed.
get_ports (optional) if true
, information about forwarded ports will be
filled in (see ports
below). This is false
by default because it may take
some time to run.
If you have multiple Vagrantfiles, provide an alias
in the provider
block
and use the provider
meta-argument in the resource/data-source
configurations.
Outputs
-
machine_names.#
- a list of machine names as defined in the Vagrantfile. -
ssh_config.#
- SSH connection info. Since a Vagrantfile may create multiple machines, this is a list with the following variables:-
ssh_config.*.type
- always "ssh" for now -
ssh_config.*.user
- the user for the connection -
ssh_config.*.host
- the address to connect to -
ssh_config.*.port
- the port to connect to -
ssh_config.*.private_key
- private ssh key for the connection -
ssh_config.*.agent
- whether or not to use the agent for authentication (always "false" for now).
If there is only one machine built by the Vagrantfile, the connection info will be set in the
resource
block so you can include provisioners without any additional configuration. However, if there is more than one machine, the connection info will not be set; you'll need to create somenull_resources
to do your provisioning. -
-
ports.#
- information about forwarded ports ifget_ports
istrue
. This is a list of lists: for each machine in the Vagrantfile,ports
will have a list with the following variables:-
ports.*.*.guest
- the port on the guest VM -
ports.*.*.host
- the host port forwarded to the guest VM
-
Note that machine_names
, ssh_config
, and ports
are guaranteed to be in
the same order (ie, ssh_config[0]
is the corresponding config for the machine
named machine_names[0]
), but the order is undefined (ie, don't count on
machine_names[0]
being the first machine defined in the Vagrantfile).
Forcing an Update
The easiest way to force an update is to set, or change the value of, some environment variable. This will signal to terraform that the vagrant_vm resource needs to update.
For example, if you want to force updates when your Vagrantfile changes, try something like this:
resource "vagrant_vm" "my_vagrant_vm" {
vagrantfile_dir = "path/to/dir"
env = {
VAGRANTFILE_HASH = md5(file("path/to/dir/Vagrantfile")),
}
}
When the file changes, the hash will change, and terraform will ask for an update.
Really Forcing an Update
Changing an environment variable, as suggested above, essentially runs vagrant reload
. Sometimes this isn't enough. If the resource's name
changes, it will
signal to terraform that it needs to completely destroy the resource and
recreate it.
Removing Machines
Sadly, due to some limitations in vagrant, it's not possible to automatically remove a portion of machines from a Vagrantfile. In other words, if your Vagrantfile defines 5 machines and you remove 2 of them from the Vagrantfile, they will be left running in your vagrant provider (ie, virtualbox or whatever) with no way of removing them via vagrant (or terraform).
If you intend of removing some machines, you should manually run vagrant destroy MACHINE_NAME
on those machines you wish to remove before editing the
Vagrantfile. Then update your Vagrantfile and allow terraform to do the rest.
If you forget, you can manually cleanup these old VMs by launching your vagrant
provider's UI and deleting the machines. Then run vagrant global-status --prune
to cleanup vagrant's cache of these machines.
Debugging
If terrafrom is failing on the vagrant step, you can get additional output by running terraform with logging output enabled. Try something like:
env TF_LOG=TRACE terraform apply ...
And, of course, you can always run vagrant on your Vagrantfile directly.
Local Development
The example in examples/resources/vagrant_vm
is fully functioning, but you'll
need to compile this provider and put it in a place terraform can find it:
go build
mkdir -p examples/resources/vagrant_vm/terraform.d/plugins/registry.terraform.io/bmatcuk/vagrant/4.0.0/darwin_amd64
mv terraform-provider-vagrant examples/resources/vagrant_vm/terraform.d/plugins/registry.terraform.io/bmatcuk/vagrant/4.0.0/darwin_amd64/
cd examples/resources/vagrant_vm
terraform init
terraform apply
Adjust darwin_amd64
to match your system.