dbterra

dbt Cloud™️ infrastructure-as-code that is friendly to analysts, bizops, etc

Motivation

While there are Terraform™️ plugins that can be used to work with dbt Cloud™️ jobs, environments, etc, we wanted a simple way to be able to create/modify/delete jobs within the same repository in a readable way that doesn't require any DevOps knowledge.

Usage

Usage: dbterra [OPTIONS] [COMMAND]

Commands:
  plan   Plans the changes derived from your dbt_cloud.yml file
  apply  Plans and applies the changes derived from your dbt_cloud.yml file
  help   Print this message or the help of the given subcommand(s)

Options:
  -d, --debug...  Turn debugging information on
  -h, --help      Print help
  -V, --version   Print version

In order to run, dbterra expects to environment variables to be set:

DBT_CLOUD_ACCOUNT_ID=123
DBT_CLOUD_TOKEN=abc123456xyz

DBT_CLOUD_ACCOUNT_ID

Your DBT_CLOUD_ACCOUNT_ID can be found by looking at the URL you use to access dbt Cloud

https://cloud.getdbt.com/deploy/<account_id>/projects/<project_id>/jobs

You may also declare this via account in your dbt_cloud.yml file.

DBT_CLOUD_TOKEN

Your DBT_CLOUD_TOKEN can be found at the bottom of your profile page

dbt_cloud.yml

The basic setup required that the dbt_cloud.yml file is present. dbterra looks for this file in the root folder of the current working directory. A sample file below demonstrates what this might look like:

# optionally allowed or can be set via DBT_CLOUD_ACCOUNT_ID 
# account:
#  id: 123

projects:
  example_project:
    # project_id in dbt cloud https://cloud.getdbt.com/deploy/<account_id>/projects/<project_id>/jobs
    id: 123
    jobs:
      seed:
        # the `name` attribute is automatically set from the "Title Case" of the YAML key
        # name: Seed
        environment: bizops
        target: production
        threads: 4
        steps:
          - dbt seed
      full_run:
        # if you want the name to be different than the YAML key, you may set it manually
        name: Full Production Run
        environment: bizops
        target: production
        steps:
          - dbt run
        generate_docs: true
        threads: 32
        schedule:
          cron: "0 9 * * *"
      
environments:
  bizops:
    # environment_id in dbt cloud https://cloud.getdbt.com/deploy/<project_id>/projects/<project_id>/environments/<environment_id>
    id: 456

Because we want to keep things simple and avoid storing state anywhere, the name is used as the unique identifier for a job within each project. This means if you rename something, it will first be deleted and then re-created.

The name key is optional and will default to the "Title Case" of the key of the job in the YAML file if not specified.

Installation

For convenience, we've built binaries for both x86_64 and aarch64 linux for both musl and gnu variants under the Releases section. If you want to install this on another system (such as Mac OS) and have cargo installed, you can use:

 cargo install dbterra --git https://github.com/Instawork/dbterra.git

Full YAML Options

account:
  id: 43811

projects:
  fishtown_analytics:
    id: 1234
    jobs:
      partial_run:
        environment: production
        target: somethingotherthandefault
        threads: 8
        defer_to_job_id: 22222
        steps:
          - dbt run --defer --select state:modified+
        generate_docs: true
      github_pr:
        name: Github PR
        environment: github
        target: gh
        threads: 4
        ci:
          run_on_pr: true
        defer_to_env_id: 1234
        steps:
          - dbt seed --select state:modified+
          - dbt run --fail-fast --full-refresh --defer --select state:modified+
          - dbt test --fail-fast --defer --select state:modified+
environments:
  production:
    id: 1234
  github:
    id: 5678

What's missing?

• Create/modify/delete environments (currently read-only)
• Create/modify/delete projects (currently read-only)
• Set id on an existing job instead of using name as unique identifier

Contributing

Please feel free to open a PR

Running tests

dbterra is relatively new and there aren't many tests, however there are enough to cover the basics.

To run the full test suite, use:

cargo test --all