Table of Contents
Preface
First, to make sure you're ready to go: ensure all Dependencies are on your system.
When testing with Docker, the following sections should be read:
If you plan to run benchmarks, please read the Benchmarking section.
To begin with aurweb's Docker infrastructure, see Getting Started.
Dependencies
The following table lists out requirements needed to run our docker-compose
services:
| Package | Distribution | Method |
|---|---|---|
| docker-compose | Arch Linux | pacman -S docker-compose |
| docker-compose | Other | https://docs.docker.com/engine/install/ |
Getting Started
The aurweb project now maintains a Dockerfile and docker-compose.yml. The Dockerfile builds a very basic image, and the docker-compose services use that image as a base.
For a complete rundown on all of the services available, see the aurweb Docker wiki page.
To run all services in the background:
$ FASTAPI_BACKEND=gunicorn FASTAPI_WORKERS=5 docker-compose up -d nginx
The FASTAPI_BACKEND and FASTAPI_WORKERS environment variables control
the backend and its number of workers we use to host the fastapi service.
For testing, we should align fastapi as close as possible with php.
Our php-fpm service uses 5 workers, so we should reflect that same
worker count in our fastapi service. Additionally, we want to use
the gunicorn backend, as it's what will be used in production.
These variables' defaults can also be configured through .env.
To stop services, including orphans:
$ docker-compose down --remove-orphans
To view service logs, where service is one of the services described in the
Docker wiki page:
$ docker-compose logs [--follow] service
Dummy Data
To test most features, the Docker instance needs some dummy data in it's database.
To create dummy data, use the schema/gendummydata.py script, which
can be configured using the following environment variables:
MAX_USERS- default: 38000
MAX_PKGS- default: 32000
OPEN_PROPOSALS- default: 15
CLOSE_PROPOSALS- default: 50
FORTUNE_FILE- default:
/usr/share/fortune/cookie
- default:
NOTE: The gendummydata.py script requires the bcrypt Python
dependency, which can be installed by pip or through the python-bcrypt
package on Arch Linux.
First, generate dummy data:
$ python schema/gendummydata.py dummy.sql
Then, import that dummy data into Docker's mariadb service:
$ mysql -h127.0.0.1 -P13306 -uaur -paur aurweb <dummy.sql
Accounts
Since you're using Docker, you have full control over the database. This opens up the ability for you to generate or register user accounts on your own.
Dummy Users
Users generated by gendummydata.py use the same value for both
username and password.
Registration
First, register your own account with a valid email on https://localhost:8444/register.
After registration, you will need to obtain the reset key from the mariadb database, since notifications are no-op:
$ mysql -h127.0.0.1 -P13306 -uaur -paur aurweb \
-e "SELECT ResetKey FROM Users WHERE Username = 'your_username'"
Browse to https://localhost:8444/passreset?resetkey={reset_key},
replacing {reset_key} with the real reset key. Here, you can set your
password as long as you provide the correct username which matches
{reset_key}.
Benchmarking
For benchmark comparisons to be fair, the fastapi service must be
configured to use a production backend (gunicorn) and 5 workers.
The php-fpm service also uses 5 workers.
To start up a Docker instance with a specific FastAPI backend and worker
count, users can provide a few environment variables to docker-compose:
$ FASTAPI_BACKEND=gunicorn FASTAPI_WORKERS=5 docker-compose up -d nginx
Git
Users can use the instance's Git repository just like they would on aur.archlinux.org. You will, of course, need to set a user's SSH public key to one which you own.
- SSH User:
aur - SSH Host:
localhost - SSH Port:
2222 - Project SSH URI:
ssh://aur@localhost:2222/project.git