Skip to content

Latest commit

 

History

History
56 lines (40 loc) · 2.45 KB

README.md

File metadata and controls

56 lines (40 loc) · 2.45 KB
Raw

VocPrez 3

A custom implementation of VocPrez for the NERC Vocabulary Service (NVS).

This version of VocPrez is a complete reimplementation of the Python API using FastAPI and is the lead instance of VocPrez v3.

Installation & Running

This version of VocPrez uses FastAPI, not Flask, as it's Python Framework. It can be run using uvicorn in a manner similar to Flask running on GUnicorn.

To install (Linux):

  1. clone the repo - https://github.com/surroundaustralia/VocPrez3
  2. (optional) create a Python virtual environment & activate it
    • ~$ python3 -m venv venv
    • ~$ source venv/bin/activate
  3. install required packages
    • ~$ pip install -r requirements.txt
  4. run API with uvicorn package
    • path here is for running within the VocPrez3/vocprez directory
    • ~$ uvicorn app:api --reload --port 5007 --log-level debug (or other port, other log level etc)

There is no test v. production difference in the running of FastAPI, as there is with Flask, so the above works for any form of deployment.

Config

  • all configuration variables are in config.py and have required default set
  • when run with uvicorn/gunicorn, PORT is usually set on the command line and HOST and SYSTEM_URI need not be changed from config.py
  • SPARQL_ENDPOINT in config.py is correct for NVS production deployment and is used by the http://nvs.surroundaustralia.com deployment and Nick's localhost testing too

Running with gunicorn

Gunicorn is run with uvicorn workers which then run the FastAPI application. This ensures multiple workers can be created as necissary and logging, stop/start handled better.

  • activate the virtual environment
    • source venv/bin/activate
  • run gunicorn with uvicorn workers
    • gunicorn -w 4 -k uvicorn.workers.UvicornWorker app:api -b 127.0.0.1:5007 -p gunicorn.pid >> out.log 2>&1 &

The system is now:

  • running with 4 workers
  • up on port 5007
  • logging, both access & error log, is now in out.log
  • the process PID is in file gunicorn.pid.

To shut down: kill `cat gunicorn.pid`

Test client

See the script api_test_client.py. It contains basic documentation within it.

Contacts

Lead Developer:

Nicholas Car
Data Systems Architect
SURROUND Australia Pty Ltd
[email protected]