3.10. gunicorn

gunicorn is an open source web server that implements the Python Web Server Gateway Interface (WSGI, see PEP 3333).
There are three connection methods between gunicorn and Nginx:

  • If Nginx is on the same system a UNIX socket will be used. Therefore a group wsgi-nginx is created and both gnuhealth & ẁwww-data have to be part of that group. Otherwise you would either have a world-writable socket or could not access your server.

  • Besides HTTP can be used by Nginx if systems are separated but the intranet is considered secure.

  • Alternatively it can also provide HTTPS for encrypted communication.

The role ships a config file gunicorn.cfg.py and a systemd service in case of GNU/Linux.

gunicorn documentation:
https://docs.gunicorn.org/en/latest/index.html

  • Parameters:

    • gu_instances: List of gunicorn instances to deploy, selections are provided in defaults

    • gu_upgrade: If true pass latest to Ansibles pip module instead of present

  • Defaults:

    • gu_ansible_comment: Put gu_template_comment on top of templates if true

    • gu_template_comment: Comment put on top of every template delivered by Ansible

    • gu_https: Spawn HTTPS server if true

    • gu_example_instance: Dictionary containing all parameters that are possibly used by the role, no working default config, just to show what can be configured. See gnuhealth and thalamus roles defaults for example configurations.

  • Variables:

    • gu_packages: Packages to be installed on the system differentiated by OS family

    • gu_venv_python: Python executable used for virtual environment

    • gu_nginx_group_name: Name of the group to be created for Nginx user & gunicorn user in order to have proper UNIX socket permissions (if necessary)

    • gu_default_mode: Contains default modes for files and directories differentiated by OS