carenage

Carenage is a continuous integration tool meant to make environmental evaluation easier for software projects. Used in CI scripts, it queries needed metrics and metadata during the execution of a CI job to insert them into a database. Afterwards, the data is available as JSON objects through requests to a RESTful API and visualisable in a dashboard.

Carenage (careening in English) is a series of operations made on the hull of a ship in order to inspect, clean or repair it. Environmental evaluation can be difficult to put into place for software development teams. With data recovered during tests or benchmarks, Carenage aims to simplify this and allow for better evaluation and better application of eco-design practices.

Using Carenage

You can use Carenage in your continuous integration scripts, if the CI environment you will be using has the needed services already set up. The section Setting up Carenage gives a bit more information on that. If that is the case, you can include the template available in ci/templates that downloads and sets up carenage for a Debian environment. This can be done with the following command in the .gitlab-ci.yml of your project:

include:
  - remote: "https://gitlab.com/hubblo/carenage/-/raw/main/ci/templates/setup-carenage.yml"
    inputs:
     lifetime: 6 

With inputs, some parameters used by carenage can be modified in place of the default ones, as documented in the template.

Self-hosting Carenage

carenage is meant to be executed in a continuous integration environment, such as a Gitlab runner. As it is dependent on several other services, namely boaviztapi and scaphandre through boagent and a PostgreSQL database, there is a bit of configuration to take care of if you want to self-host carenage.

Gitlab runner

If you have set up your own Gitlab runner, you can add all needed services so that any CI pipeline can be executed without having to repeat each service setup process for each job. In the folder ci/docker, you can find examples of a minimal config.toml for a Gitlab runner, and of a docker-compose.yml to see how to orchestrate all needed containers, with the associated network and volumes. You may have to tweak the configuration for your own specific needs, but with this minimal configuration and orchestration set into place, a CI script similar to ci/templates/setup-carenage.yml should be able to use carenage.

Development

Back-end

The carenage project is divided into several packages, with three binaries (api, carenaged, carenage-cli) and one library (database).

• carenaged contains the daemon program that inserts metadata and metrics into the database.
• carenage-cli contains the command-line interface program that can start and stop the execution of the daemon inside scripts.
• api contains the web server that receives requests and sends responses through a RESTful API.

The database library contains all methods needed to communicate with the database, most data structures used throughout the project, and modules to communicate with other external services (most notably boagent).

To create environmental impacts metrics, carenage queries boagent to receive data on the hardware for the computer running processes. This data is made of environmental impact metrics (through queries to boaviztapi), and energy consumption metrics (through scaphandre). carenage queries information on the computer configuration, then, throughout a CI script execution, queries information for each process running on the computer. All these pieces of information are then inserted into the database with a timestamp.

Building the project

With rust and cargo installed:

cd carenage
make build_debug

Testing

Unit and integrations tests are implemented for the different packages of the project:

cd carenage
make test

Some tests need a Postgres database set up to apply migrations and fixtures. If you have Docker installed, make compose_dev will set up the development / testing environment.