Chess-cloud is a prototype system for analysing chess posisitons in cloud environment using StockFish5 chess engine (oen of the best chess eninges that is also open-sourced). Using Akka Cluster, work (chess analysis) is distributed among available workers runnign on different computing nodes. As a foundation pattern described in blog post Balancing Workload Across Nodes with Akka 2 was used.
Chess cluster is accessible through web application writen in AngularJS and PlayFramework. Frontend provides simple and interactive user interface, which allows
- Playing Chess game with Computer (Cluster Engine)
- Analysing chess games
- supports PGN (Portable Game Notation)
- Play-out current position
- Small DB of one-of-the-best chess games ever played
Running instance can be found at Chess-cloud.com.
Frontend: cd frontend/ && sbt run
Cluster: cd cluster/ && sbt run
Open localhost:9000
This commands runs frontend web application and cluster in a development mode. Cluster is made of 1 master available on localhost:2551, 2 workers on random ports and a frontend that is run as part of frontend application in separate actor system.
Deployment is done using Fabric scripts and LibCloud. Atm, scirpts supports AWS and OpenStack, however adding additional provider should be fairly easy (extending scripts/driver.py).
Python, VirtualEnv, Pip > Set up python virtual environment and install requirements.
chesscloud-deps image available in target cloud enviornment. It can be default image, where
provision is run (installs Java8) on created nodes.
Configuration file conf.ini holds cloud provider authentication, target infrastructure properties (VM sizes), and
name of prepared image chesscloud-deps, which should be prepared in advance and will be used when spwning new VMs.
# Creating and deploying cluster
fab create_cluster
fab deploy_cluster
# Scaling up and down (argument: cluster_node=[frontend|master|worker])
fab scale_up
fab scale_down
# Listing nodes - or selection
fab frontends | masters | workers
fab cluster | all
# Starting and stopping cluster or selected nodes
fab stop_cluster | start_cluster
fab <selection> start|stop
# Destroying cluster or selected nodes
fab destroy_cluster
fab <slection> destroy
# Misc commands - ssh to machine, execute cmd on target machine, ...
fab <selection> ssh | cmd get_log | tail_log