ansible | ||
bin | ||
docs | ||
feedler | ||
mockup | ||
publichealth | ||
tests | ||
vagrant | ||
.bowerrc | ||
.buildpacks | ||
.dockerfile | ||
.editorconfig | ||
.gitignore | ||
ansible.cfg | ||
bitbucket-pipelines.yml | ||
bower.json | ||
CHANGELOG.md | ||
fabfile.py | ||
Gruntfile.js | ||
LICENSE.md | ||
Makefile | ||
manage.py | ||
package.json | ||
Procfile | ||
README.md | ||
requirements-test.txt | ||
requirements.txt | ||
runtime.txt | ||
stellar.yaml | ||
Vagrantfile |
Public Health Schweiz
Website of the Swiss Society for Public Health, developed by datalets,ch using the open source, Django-based Wagtail CMS. The frontend is implemented by moving water using Bootstrap framework.
This project is open source under the MIT License.
Development environment
The easiest way to set up your machine would be to use Vagrant, then in the project folder in the terminal type: vagrant up
. Then when it is ready, follow instructions for Database setup.
To set up a full development environment, follow all these instructions.
Frontend setup
If not using Vagrant, you will need to have Ruby and SASS installed on your system, e.g.:
sudo apt-get install ruby-sass
Make sure a recent version of node.js (we recommend using nave.sh), then:
npm install -g bower grunt-cli
npm install
bower install
The first command (..install -g..
) may require sudo
if you installed node.js as a system package.
If you are only working on the frontend, you can start a local webserver and work on frontend assets without the backend setup described below. Mock content is at mockup
, and there is a grunt browser-sync
setup for working with frontend assets.
Backend setup
If not using Vagrant: after installing Python 3, from the project folder, deploy system packages and create a virtual environment as detailed (for Ubuntu users) below:
sudo apt-get install python3-venv python3-dev libjpeg-dev
pyvenv env
. env/bin/activate
pip install -U pip
pip install -r requirements.txt
At this point your backup is ready to be deployed.
Database setup
Once your installation is ready, you can get a blank database set up and add a user to login with.
If you are using Vagrant, enter the shell of your virtual machine now with vagrant ssh
Run these commands:
./manage.py migrate
./manage.py createsuperuser
You will be asked a few questions to create an administrator account.
Starting up
If you have one installed, also start your local redis server (service redis start
).
After completing setup, you can use:
./manage.py runserver
(In a Vagrant shell, just use djrun
)
Now access the admin panel with the user account you created earlier: http://localhost:8000/admin/
Troubleshooting
- Issues with migrating database tables in SQLite during development? Try
./manage.py migrate --fake
Production notes
We use Ansible and Docker Compose for automated deployment.
To use Docker Compose to deploy the site, copy ansible/roles/web/templates/docker-compose.j2
to /docker-compose.yml
and fill in all {{ variables }}
. This is done automatically in Ansible.
To do production deployments, you need to obtain SSH and vault keys from your system administrator (who has followed the Ansible guide to set up a vault..), and place these in a .keys
folder. To deploy a site:
ansible-playbook -s ansible/<*.yaml> -i ansible/inventories/production
For an update release with a specific version, use:
ansible-playbook -s ansible/site.yaml -i ansible/inventories/production --tags release -e gitversion=<v*.*.*>
We use a StackScript to deploy to Linode, the basic system set up is to have a user in the sudoers and docker group, and a few basic system packages ready.
For example, on Ubuntu:
apt-get install -q -y zip git nginx python-virtualenv python-dev
The order of deployment is:
- docker.yaml (base system)
- node.yaml
- site.yaml
- harden.yaml
- certbot.yaml
The last line adds support for Let's Encrypt, which you can configure and enable (updating your Nginx setup) with:
sudo /opt/certbot/certbot-auto --nginx certonly
If you do not wish to use SSL, delete the last part of your nginx site configuration (/etc/nginx/sites-enabled/...).
Production releases
For further deployment and system maintenance we have a Makefile
which automates Docker Compose tasks. This should be converted to use Ansible Container. In the meantime, start a release with Ansible, then complete it using make
, i.e.:
ansible-playbook -s ansible/site.yaml -i ansible/inventories/production --tags release
ssh -i .keys/ansible.pem ansible@<server-ip> "cd <release_dir> && make release"
Restoring a data backup
For development, it's handy to have access to a copy of the production data. To delete your local database and restore from a file backup, run:
rm publichealth-dev.sqlite3
python manage.py migrate
python manage.py loaddata publichealth.home.json
You might want to createsuperuser
again at this point.