This is a pre-production deployment of Warehouse, however changes made here WILL affect the production instance of PyPI.
Latest Version Dependencies status unknown Test status unknown Test coverage unknown
Project Description

This is a demo implementation of the Paxos algorithm implemented in Python.


  • Python 3


pip install paxos

Implementation Notes

  • The proposer, acceptor, and learner roles of the Paxos algorithm are implemented in with classes that subclass from a common Agent class.
  • Each role/agent is run in a separate process.
  • Communication between processes occurs using Queue objects, so all processes are run on the same machine.
  • Paxos Made Simple states that “we require that different proposals have different numbers.” To achieve this, we start each proposer process’s proposal number sequence equal to its own PID, and then increment the number for each new proposal by the number of proposer processes in the system. This also seems to be the method used in the “Paxos Made Live” paper by Google employees.
  • It is assumed that all processes in the system be considered members of the system from the beginning, without needing to explicitly join the system by getting a decree passed.



  • Add a collapsed version of the roles so that each process plays all of the roles.
    • Once we have a collapsed version, leaders should retry a client’s request if they determine that the instance hasn’t been decided after some timeout amount of time. This should fix a couple issues:
      1. Learners are not able to determine whether or not there are more values to learn (when it is the last value they are missing).
      2. Since leaders are currently remembering the original value of each client request they propose, if a Proposer is asked to retry an instance (e.g. from a Learner that is missing a value) then it is possible that the value learned will be None in the situation where no Acceptor has yet accepted a value in that instance (which means the Proposer should specify the value, but since it is not remembering the original values it just proposes None).
  • When a learner asks a proposer to retry, the proposer shouldn’t retry if it has already retried that proposal within a certain time period because otherwise, by re-upping the proposal number it would be guaranteed to not have a successful agreement in that instance it is retrying.
  • With a consistent leader, only perform phase one once.
Release History

Release History


This version

History Node

TODO: Figure out how to actually get changelog content.

Changelog content for this version goes here.

Donec et mollis dolor. Praesent et diam eget libero egestas mattis sit amet vitae augue. Nam tincidunt congue enim, ut porta lorem lacinia consectetur. Donec ut libero sed arcu vehicula ultricies a non tortor. Lorem ipsum dolor sit amet, consectetur adipiscing elit.

Show More

Download Files

Download Files

TODO: Brief introduction on what you do with files - including link to relevant help section.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
paxos-0.1.tar.gz (16.8 kB) Copy SHA256 Checksum SHA256 Source Jul 28, 2012

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting