Deployment

While the default development server (flask) is fine for running kmviz locally, it is recommended to use a production-ready server to deploy an instance that can be accessed by several users.

kmviz offers a small CLI tool to deploy an instance using gunicorn.

python -m kmviz app deploy --help
Usage: python -m kmviz app deploy [OPTIONS]

Deployment with gunicorn (recommended for multi-user instances)

Options:
-m, --mode <str> Mode, 'db' or 'api' [default: db]
-u, --url <str> Host [default: localhost]
-p, --port <int> Port [default: 8000]
-w, --workers <int> Number of workers [default: 25]
-r, --reload-extra-file <str> Reload when files change
-e, --extra <option> <value> Extra gunicorn options
--help Show this message and exit.

Ex: python -m kmviz app -c config.toml deploy -u 0.0.0.0 --port 8080 -w 8

We provide a docker image, tlemane/kmviz, to easily deploy kmviz using gunicorn (see Docker).

See Docker compose for a complete deployment example including:

• A sequence index, using kmindex.
• A metadata db, using MySQL.
• A kmviz instance

---

Deploy with another server

The following WSGI servers have been tested:

• gunicorn
• fastwsgi
• socketify

Note

The WSGI application is located at kmviz/app.py and the variable named is app. For example with gunicorn: gunicorn -w 1 kmviz.app:app

Note

When executed with a WSGI server, the path to the configuration file is given by the KMVIZ_CONF environment variable.

---