Skip to content

Latest commit

 

History

History
150 lines (113 loc) · 3.8 KB

INSTALL.md

File metadata and controls

150 lines (113 loc) · 3.8 KB
Raw

Radar

This document describes installation for production use.

System

As root

  1. Install required packages

    apt install \
    python3-venv \
    python3-dev \
    memcached \
    libmemcached11 \
    zlib1g \
    rabbitmq-server \
    nginx \
    postgresql \
    libpq-dev \
    build-essential

    libmemcached11 and zlib1g are needed for the pylibmc backend for Memcached.

  2. Create a new user for radar

    adduser --system --group \
    --shell /bin/bash --home /srv/radar \
    --gecos "A-plus radar service" \
    radar

  3. Create a database and add permissions

    sudo -Hu postgres createuser radar
sudo -Hu postgres createdb -O radar radar

  4. Create a run directory

    echo "d /run/radar 0750 radar www-data - -" \
    | tee /etc/tmpfiles.d/radar.conf > /dev/null
systemd-tmpfiles --create

Django

  1. Change to the service user

    su - radar

  2. Clone the application

    # Run as user radar in /srv/radar
python3 -m venv venv
source ~/venv/bin/activate
git clone https://github.com/apluslms/radar.git
cd radar
# Check out to whichever version is desired
git checkout v0.0
pip install -r requirements-prod.txt
pip install -r requirements.txt

cp radar/local_settings.example.py radar/local_settings.py
# Add correct hostname and admin email address to settings
editor radar/local_settings.py

# The secret key will be automatically generated if
# radar/secret_key.py is missing

python manage.py migrate
python manage.py collectstatic

    See README.rst for more detailed instructions about setting up A+

    python manage.py add_lti_key --desc aplus

    Select course -> Edit course -> Menu -> Add -> Select radar

    Click radar link in menu => Creates course in radar

uWSGI, NGINX and Celery workers

As root

cd ~radar/radar

  1. Install uWSGI configuration

    cp ~radar/radar/doc/uwsgi-radar.ini ~radar/uwsgi-radar.ini
chown radar:radar ~radar/uwsgi-radar.ini
cp ~radar/radar/doc/radar-web-uwsgi.service /etc/systemd/system
systemctl daemon-reload
systemctl enable radar-web-uwsgi
systemctl start radar-web-uwsgi

  2. NGINX configuration

    apt-get install nginx
# Set custom name here if $(hostname) is not correct
name=$(hostname)
sed -e "s/__HOSTNAME__/$name/g" ~radar/radar/doc/radar-nginx.conf > \
    /etc/nginx/sites-available/$name.conf
ln -s ../sites-available/$name.conf /etc/nginx/sites-enabled/$name.conf
# Edit /etc/nginx/sites-available/$name.conf if necessary (set TLS
# certificate path)
editor /etc/nginx/sites-available/$name.conf
systemctl reload nginx

  3. Install celery workers

    Worker radar-celery_matcher can also be run inside Kubernetes. This requires some extra configuration, including editing the queue name in radar/local_settings.py.

    cp ~radar/radar/doc/celery-systemd/radar-celery{,_db,_io,_matcher}.service \
    /etc/systemd/system
systemctl daemon-reload
systemctl enable radar-celery{,_db,_io,_matcher}
systemctl start radar-celery{,_db,_io,_matcher}

RabbitMQ configuration

The RabbitMQ messages between Radar and the matchlib can become large when there are several submissions to be analaysed. If the message size exceeds the default maximum message size (128 MB), it will be dropped and the Radar task never completes. Therefore, it is recommended to increase the RabbitMQ maximum message size, by adding the following line to /etc/rabbitmq/rabbitmq.conf (for doubling the max. message size to 256 MB):

max_message_size = 268435456