Distributed Fault Tolerance | Project - Phase 2 | 2024/2025

Blockchain implementation following a Streamlet consensus algorithm

This project is being developed for the Distributed Fault Tolerance course of the Master's degree in Computer Science at the Faculty of Sciences of the University of Lisbon.

Authors

Group 12:

• Eduardo Proença nº57551
• Pedro Cardoso nº58212
• Ricardo Costa nº64371

Overview

This repository contains the code of a blockchain implementation following a Streamlet consensus algorithm.

Source code organization

• domain: This directory contains data structures used to maintain the blockchain
• utils: This directory contains utility functions and scripts
• node.py: Node class that represents a node in the network
• main.py: Main script that launches the nodes

Limitations

This implementation does not write the blockchain to disk so, in the event of a crash, when the node is recovered, despite being able to catch up to the other nodes, it will not have the blocks prior to the crash. This could be further improved by writing the blockchain to disk, load it on startup and ask for the missing blocks to the other nodes from the epochs while it was down.

Requirements

• Python 3.12 or higher installed.
• PyYAML (pip install pyyaml)

Usage

To run the blockchain, you need to execute the main.py script: (in the src folder)

python main.py

This script will read the configuration file config.yaml, automatically set a time for the nodes to start and launch the nodes each in a separate terminal window.

When it hits the start time, the nodes will initiate the protocol.

Then, you can look at the node terminals to see the blockchain being built and finalized.

An execution example is shown below:

$ python main.py
Configuration file loaded
Start time set to 2024-12-08 19:51:11.368839
Node 0 started on terminal 127.0.0.1:8000
Node 1 started on terminal 127.0.0.1:8001
Node 2 started on terminal 127.0.0.1:8002

If you choose to stop a node (crash simulation), you can restart it by running:

python node.py --id <id>

Notes:

• There needs to be always a majority of nodes running for the protocol to function properly.
• Synchronization of nodes is guaranteed by the message processing thread, which only processes messages from the current or previous epochs. This means that a message from a future epoch will never be delivered in the current epoch, which ensures synchronization between nodes if their epochs are not perfectly synchronized.