Add option to use cache during building

Dockerfile

@@ -7,6 +7,7 @@ MAINTAINER Bart Cox <[email protected]>
 # Run build without interactive dialogue
 ARG DEBIAN_FRONTEND=noninteractive
 
+# Set required environmental variables for the working setup.
 ENV GLOO_SOCKET_IFNAME=eth0
 ENV TP_SOCKET_IFNAME=eth0
 
@@ -19,7 +20,9 @@ RUN apt-get update \
 
 # Copy the current folder to the working directory
 ADD setup.py requirements.txt ./
-RUN python3 -m pip install -r requirements.txt
+
+# Use cache for pip, otherwise we repeatedly pull from repository
+RUN --mount=type=cache,target=/root/.cache/pip python3 -m pip install -r requirements.txt
 
 ADD configs configs
 
@@ -28,13 +31,5 @@ ADD fltk fltk
 # Install newest version of library
 RUN python3 -m setup install
 
-
-
 # Expose the container's port to the host OS
 EXPOSE 5000
-
-# Run command by default for the executing container
-# CMD ["python3", "/opt/Generatrix/rpc_parameter_server.py", "--world_size=2", "--rank=0", "--master_addr=192.168.144.2"]
-
-#CMD python3 /opt/federation-lab/rpc_parameter_server.py --world_size=$WORLD_SIZE --rank=$RANK --master_addr=10.5.0.11
-CMD python3 /opt/federation-lab/federated_learning.py $RANK $WORLD_SIZE 10.5.0.11

README.md

@@ -8,6 +8,14 @@ Pytorch Distributed ([docs](https://pytorch.org/tutorials/beginner/dist_overview
 The goal if this project is to launch Federated Learning nodes in truly distribution fashion.
 
 This project is tested with Ubuntu 20.04 and python {3.7, 3.8}.
+
+### Building locally
+
+To build locally, run the following command in the project root directory.
+
+```bash 
+DOCKER_BUILDKIT=1 docker build . --tag fltk
+```
 ### Global idea
 Pytorch distributed works based on a world_size and ranks. The ranks should be between 0 and world_size-1.
 Generally, the federator has rank 0 and the clients have ranks between 1 and world_size-1.