Skip to content

Latest commit

 

History

History
80 lines (59 loc) · 3.69 KB

RUNNING_LOCALLY.md

File metadata and controls

80 lines (59 loc) · 3.69 KB
Raw

Below includes the documentation of how to set up and run the platform (cars) locally, this includes building the language images and setting up the loader and runner. Steps to follow when making changes and how to test execute tests.

Prerequisite

Application Version
Docker v20.10.x
Docker Compose v1.29.x
Golang v1.18.x
Make Modern
modd https://github.com/cortesi/modd

Setup

Running the application is broken up into three parts, building the language containers, running the local queue services and database setup, finally setting up and running the API and loader.

Binary Description
API The API is the consumer-facing application that does input validation and puts the content into the queue.
Loader The loader is the application that sets up the containers and communicates with docker directly to run the code.
Runner The runner is what is executed inside the container, this is what enforces limits, runs, and compiles the code.

Building the language containers

Building the language containers is the first step in running the application locally. Each supported language has its supporting image which is run on user code execution. Each image contains a copy of the runner binary which means making changes to the runner will require re-creating all the language contains (some workaround for this).

No language is required to be installed directly on the host machine for this application to work. All are done via docker images.

# build all images for all supported languages
make build-languages

# Build all images for all supported languages with extra logging
make build-languages/verbose

# Build a single image for a single language 
# This is easier for local runner development.
make build-languages CLANG=rust

Local database and queue starting

CARS requires a local database and queue to work, the application supports setting up a database and queue via a docker image. This can be done purely via a single docker-compose command. This will set up the queue and database. This must be in the root of the directory.

# start the database and queue
docker-compose up -d

# stop the database and queue
docker-compose down

Running the API and loader

Finally, the loader and the API should be executed outside the docker-compose set up, this allows faster development and turn around since the loader cannot be executed within docker as it has to communicate with the docker engine.

First, ensure modd is installed, this is what will be used to restart the application on start.

go install github.com/cortesi/modd/cmd/modd

Next, execute modd in the root directly after the database and queue setup. If all is completed correctly then INF listening on :8080 will be outputted into the console. You should now be able to access http://localhost:8080 in the browser to view the API sample site and run an execution to validate the setup.