CANOSP-2019

This project implements a minimal server that implements federated learning and accepts messages from clients.

Setting up the test environment

Install Python 3.6 or higher
Setup your virtual environment python3 -m venv venv
Install your dependencies pip install -r requirements.txt

Building a release

python setup.py sdist

Build Requirements

docker-ce : https://docs.docker.com/install/

Build Instructions

Once you have docker-ce installed, you should be able to build this project using either GNUMake or you can use docker directly.

Build the docker image using:

docker build . -t mozfldp:latest

Running the server

You can run the server locally serving requests on port 8000 using:

python -m mozfldp.server

Alternately, you can run the service in a container using :

	docker run -dit -p 127.0.0.1:8090:8000 --name mozfldp -t --rm mozfldp:latest -m mozfldp.server

Note that in the above command, we are exposing the container's port 8000 by binding it to port 8090 on the host computer.

Sending data to the server

You can submit arbitrary JSON blobs to the server using HTTP POST.

A sample curl invocation that will work is:

Note the port 8090 below.

If you are running native locally -the port will be 8000. Port 8090 is used if you are running in a docker container.

curl -d '{"key1":"value1", "key2":"value2"}' \
   -H "Content-Type: application/json" \
   -X POST http://127.0.0.1:8090/api/v1/client_update/some_client_id

Client ID: [some_client_id].
JSON Payload: {'key1': 'value1', 'key2': 'value2'}%