Configuring a Remote Workspace

When you run rx init, rx sets up a “workspace,” which is a Docker container on a remote host. You have a lot of control over how this container is set up and configured.

The Config File

rx determines the type of container and hardware to use remotely based on the contents of a YAML file. If you do not specify a path to a YAML file, rx defaults to using the one .rx/remotes/default is symlinked to. If this does not exist, rx will attempt to detect the tools and languages the project is using. If it cannot, it will fallback on a basic machine with a recent version of Python installed.

To modify this behavior you can use a different config file, either by changing the target the default symlink points to or by specifying a path on the command line:

$ rx init --remote=some/path/to/my/config.yaml

Once a workspace is initialized the remote config file is not re-read.

A config file has two sections:

  • image: describes the Docker container to start.
    • registry: Optional, defaults to Docker Hub.
    • repository: The Docker image name.
    • tag: Defaults to latest.
    • envrionment_variables: Key/value pairs of environment variables to set for the workspace.
  • remote: describes the machine the workspace should be put on.
    • toolchain: which tools your project is using.
    • hardware: The CPU, memory, and disk requirements.
      • processor: The type of processor needed.

Using an Existing Config

If a project has an existing rx config then you can use it to quickly initialize a workspace. For example, suppose you have cloned an open source project with an rx.yaml file. You can run:

$ rx init --remote=rx.yaml

This will set up a workspace in the cloud with the config specified in the rx.yaml file.

Modifying a Workspace

As you work on a project you may install tools, edit configs, or generate artifacts on the remote machine. If you’d like to save that state, you can run:

$ rx ws commit
Storing your workspace...
Pushed layer cb9317ff7c32: : 23501824 bytes [00:00, 32461083.90 bytes/s]
Your remote machine's state has been saved.

If you'd like to initialize a new workspace with this state, use the following lines in your config:

  repository: username/rx-root-dir
  tag: '20240120'

This will save the current state of your workspace. (Note that this is not useful for saving the current state of your project, since that is synced from your local device every time you run a command.)

The output of the commit command prints the stanza you’d need to put in your config to init a remote workspace from this shapshot. You can also always get this infomation by running rx ws info.

Sharing Configs

By default, your user is the only one who access to the workspace images you create. However, you may want to share your image with others. Sharing your image allows others to start up their own workspace with the given state, but does not allow them to write new versions of it to your repository. They can, however, write their own version to their own repository.

To share your image with a user (in this case, “trexbot”), run:

$ rx ws set-acls --add-reader=trexbot

If you are using rx with an enterprise plan you can also share images with groups that you are a member of:

$ rx ws set-acls
$ # or a subgroup:
$ rx ws set-acls

You must be a member of a group to share it with them.

For open source projects, you may want to give anyone access to use a pre-configured environment. You can do this by setting the default visibility to public:

$ rx ws set-acls --visibility=public

If you change your mind you can set the visibility back to “private” (the default).

Changing the Docker Image

Machines can be configured to use any public Docker image, as specified in the image.repository field. For example, if you wanted a plain Ubuntu machine you could create a file /path/to/rxroot/my-rx-remotes/vanilla-ubuntu.yaml:

  repository: ubuntu
  tag: "22.04"

Then run rx init using the --remote option:

$ rx init --remote=my-rx-remotes/vanilla-ubuntu.yaml


You can use any publicly-available Docker image. If you’d like to use a private image, please let us know.

At the moment, there is a cap of 8GB on image size. Again, please email us if you’d like to use a larger image.