» Command: init
Hands-on: Try the Terraform: Get Started collection on HashiCorp Learn.
terraform init command is used to initialize a working directory
containing Terraform configuration files. This is the first command that should
be run after writing a new Terraform configuration or cloning an existing one
from version control. It is safe to run this command multiple times.
terraform init [options]
This command performs several different initialization steps in order to prepare the current working directory for use with Terraform. More details on these are in the sections below, but in most cases it is not necessary to worry about these individual steps.
This command is always safe to run multiple times, to bring the working directory up to date with changes in the configuration. Though subsequent runs may give errors, this command will never delete your existing configuration or state.
» General Options
The following options apply to all of (or several of) the initialization steps:
-input=trueAsk for input if necessary. If false, will error if input was required.
-lock=falseDisable locking of state files during state-related operations.
-lock-timeout=<duration>Override the time Terraform will wait to acquire a state lock. The default is
0s(zero seconds), which causes immediate failure if the lock is already held by another process.
-no-colorDisable color codes in the command output.
-upgradeOpt to upgrade modules and plugins as part of their respective installation steps. See the sections below for more details.
» Copy a Source Module
terraform init assumes that the working directory already
contains a configuration and will attempt to initialize that configuration.
Optionally, init can be run against an empty directory with the
-from-module=MODULE-SOURCE option, in which case the given module will be
copied into the target directory before any other initialization steps are
This special mode of operation supports two use-cases:
Given a version control source, it can serve as a shorthand for checking out a configuration from version control and then initializing the working directory for it.
If the source refers to an example configuration, it can be copied into a local directory to be used as a basis for a new configuration.
For routine use it is recommended to check out configuration from version
control separately, using the version control system's own commands. This way
it is possible to pass extra flags to the version control system when necessary,
and to perform other preparation steps (such as configuration generation, or
activating credentials) before running
» Backend Initialization
During init, the root configuration directory is consulted for backend configuration and the chosen backend is initialized using the given configuration settings.
Re-running init with an already-initialized backend will update the working
directory to use the new backend settings. Depending on what changed, this
may result in interactive prompts to confirm migration of workspace states.
-force-copy option suppresses these prompts and answers "yes" to the
migration questions. The
-reconfigure option disregards any existing
configuration, preventing migration of any existing state.
To skip backend configuration, use
-backend=false. Note that some other init
steps require an initialized backend, so it is recommended to use this flag only
when the working directory was already previously initialized for a particular
-backend-config=... option can be used for
partial backend configuration,
in situations where the backend settings are dynamic or sensitive and so cannot
be statically specified in the configuration file.
» Child Module Installation
During init, the configuration is searched for
module blocks, and the source
code for referenced modules is retrieved from the locations
given in their
Re-running init with modules already installed will install the sources for
any modules that were added to configuration since the last init, but will not
change any already-installed modules. Use
-upgrade to override this behavior,
updating all modules to the latest available source code.
To skip child module installation, use
-get=false. Note that some other init
steps can complete only when the module tree is complete, so it's recommended
to use this flag only when the working directory was already previously
initialized with its child modules.
» Plugin Installation
Most Terraform providers are published separately from Terraform as plugins. During init, Terraform searches the configuration for both direct and indirect references to providers and attempts to install the plugins for those providers.
For providers that are published in either
the public Terraform Registry or in a
third-party provider registry,
terraform init will automatically find,
download, and install the necessary provider plugins. If you cannot or do not
wish to install providers from their origin registries, you can customize how
Terraform installs providers using
the provider installation settings in the CLI configuration.
For more information about specifying which providers are required for each of your modules, see Provider Requirements.
After successful installation, Terraform writes information about the selected
providers to the dependency lock file.
You should commit this file to your version control system to ensure that
when you run
terraform init again in future Terraform will select exactly
the same provider versions. Use the
-upgrade option if you want Terraform
to ignore the dependency lock file and consider installing newer versions.
You can modify
terraform init's plugin behavior with the following options:
-upgradeUpgrade all previously-selected plugins to the newest version that complies with the configuration's version constraints. This will cause Terraform to ignore any selections recorded in the dependency lock file, and to take the newest available version matching the configured version constraints.
-get-plugins=false— Skip plugin installation.
-plugin-dir=PATH— Force plugin installation to read plugins only from the specified directory, as if it had been configured as a
filesystem_mirrorin the CLI configuration. If you intend to routinely use a particular filesystem mirror then we recommend configuring Terraform's installation methods globally. You can use
-plugin-diras a one-time override for exceptional situations, such as if you are testing a local build of a provider plugin you are currently developing.
terraform init in automation
For teams that use Terraform as a key part of a change management and deployment pipeline, it can be desirable to orchestrate Terraform runs in some sort of automation in order to ensure consistency between runs, and provide other interesting features such as integration with version control hooks.
There are some special concerns when running
init in such an environment,
including optionally making plugins available locally to avoid repeated
re-installation. For more information, see
the Running Terraform in Automation tutorial on HashiCorp Learn.
» Passing a Different Configuration Directory
Terraform v0.13 and earlier also accepted a directory path in place of the
plan file argument to
terraform apply, in which case Terraform would use
that directory as the root module instead of the current working directory.
That usage is still supported in Terraform v0.14, but is now deprecated and we
plan to remove it in Terraform v0.15. If your workflow relies on overriding
the root module directory, use
-chdir global option
instead, which works across all commands and makes Terraform consistently look
in the given directory for all files it would normaly read or write in the
current working directory.
If your previous use of this legacy pattern was also relying on Terraform
.terraform subdirectory into the current working directory even
though the root module directory was overridden, use
TF_DATA_DIR environment variable
to direct Terraform to write the
.terraform directory to a location other
than the current working directory.