Skip to main content

Provisioning Infrastructure with OpenTofu

OpenTofu's primary function is to create, modify, and destroy infrastructure resources to match the desired state described in a OpenTofu configuration.

When people refer to "running OpenTofu," they generally mean performing these provisioning actions in order to affect real infrastructure objects. The OpenTofu binary has many other subcommands for a wide variety of administrative actions, but these basic provisioning tasks are the core of OpenTofu.

OpenTofu's provisioning workflow relies on three commands: plan, apply, and destroy. All of these commands require an initialized working directory, and all of them act only upon the currently selected workspace.


The tofu plan command evaluates a OpenTofu configuration to determine the desired state of all the resources it declares, then compares that desired state to the real infrastructure objects being managed with the current working directory and workspace. It uses state data to determine which real objects correspond to which declared resources, and checks the current state of each resource using the relevant infrastructure provider's API.

Once it has determined the difference between the current state and the desired state, tofu plan presents a description of the changes necessary to achieve the desired state. It does not perform any actual changes to real world infrastructure objects; it only presents a plan for making changes.

Plans are usually run to validate configuration changes and confirm that the resulting actions are as expected. However, tofu plan can also save its plan as a runnable artifact, which tofu apply can use to carry out those exact changes.

For details, see the tofu plan command.


The tofu apply command performs a plan just like tofu plan does, but then actually carries out the planned changes to each resource using the relevant infrastructure provider's API. It asks for confirmation from the user before making any changes, unless it was explicitly told to skip approval.

By default, tofu apply performs a fresh plan right before applying changes, and displays the plan to the user when asking for confirmation. However, it can also accept a plan file produced by tofu plan in lieu of running a new plan. You can use this to reliably perform an exact set of pre-approved changes, even if the configuration or the state of the real infrastructure has changed in the minutes since the original plan was created.

For details, see the tofu apply command.


The tofu destroy command destroys all of the resources being managed by the current working directory and workspace, using state data to determine which real world objects correspond to managed resources. Like tofu apply, it asks for confirmation before proceeding.

A destroy behaves exactly like deleting every resource from the configuration and then running an apply, except that it doesn't require editing the configuration. This is more convenient if you intend to provision similar resources at a later date.

For details, see the tofu destroy command.