Codabench Installation

Installation Guide

Installing Codabench should be relatively easy since you no longer have to worry about special ways to setup SSL or storage. We include default solutions that should handle that for most basic uses.

1. Pre-requisites

Install Docker

Either:

a) Install docker via the install script: https://docs.docker.com/install/linux/docker-ce/ubuntu/#install-using-the-convenience-script

b) Install manually following the steps at: https://docs.docker.com/install/

Install Docker-Compose

https://docs.docker.com/compose/install/

2. Clone Repository

Download the Codabench repository:

git clone https://github.com/codalab/codabench

3. Edit the settings (.env)

The .env file contains the settings of your instance. At first you don't have the .env file. To create it, copy the .env_sample as following:

cd codabench
cp .env_sample .env

Then edit the necessary settings inside. The most important are the database, storage, and Caddy/SSL settings. For a quick local setup, you should not need to edit this file.

If you are using AWS_S3_ENDPOINT_URL=http://minio:9000/ in your .env, edit your /etc/hosts file by adding this line 127.0.0.1 minio

For MacOS

In .env, replace:

AWS_S3_ENDPOINT_URL=http://minio:9000/

by

AWS_S3_ENDPOINT_URL=http://docker.for.mac.localhost:9000/

If needed, some troubleshooting of this step is provided at the end of this page or in this page.

4. Run the following commands

Create the required tables in the database:

docker-compose exec django ./manage.py migrate

Generate the required static resource files:

docker-compose exec django ./manage.py collectstatic --noinput

5. Start the service

To deploy the platform, run:

docker-compose up -d

You should be able to verify it is running correctly by looking at the logs with docker-compose logs -f and by visting localhost:80 (Depending on your configuration).

Advanced Configuration

Testing

To run automated tests for your local instance, get inside the Django constainer with docker-compose exec django bash then run py.test to start the automated tests.

SSL

To enable SSL:

• If you already have a DNS for your server that is appropriate, in the .env simply set DOMAIN_NAME to your DNS. Remove any port designation like :80. This will have Caddy serve both HTTP and HTTPS.

Validate user account on local instance

When deploying a local instance, the email server is not configured by default, so you won't receive the confirmation email during signup.

To manually confirm your account:

1. Find the confirmation link in the Django logs using docker-compose logs -f django
2. Replace example.com by localhoston the URL and open it on the browser.

Another way is to go inside the Django containers and use commands like in administrator procedures.

Troubleshooting storage endpoint URL

You may have to manually change the endpoint URL to have your local instance working. This may be an OS related issue. Here is a possible fix:

1. docker-compose logs -f minio
2. Grab first of these:

minio_1           | Browser Access:
minio_1           |    http://172.27.0.5:9000  http://127.0.0.1:9000

3. Set AWS_S3_ENDPOINT_URL=http://172.27.0.5:9000in your .env file.

For Apple CPU (M1, M2 chips)

In docker-compose.yml, replace in the compute_worker service:

command: bash -c "watchmedo auto-restart -p '*.py' --recursive -- celery -A compute_worker worker -l info -Q compute-worker -n compute-worker@%n"

by

command: bash -c "celery -A compute_worker worker -l info -Q compute-worker -n compute-worker@%n"

Storage

By default v2 uses a built-in minio container. Some users may want a different solution such as S3 or Azure. The configuration will vary slightly for each different type of storage.

For all possible supported storage solutions, see: https://django-storages.readthedocs.io/en/latest/

Remote Compute Workers

For setting up remote compute workers, please see this link to the wiki article on setting up compute workers: Compute Worker Management

Troubleshooting

Read the following guide for troubleshooting: How to deploy Codabench.