elaine 0.0.3

data/lib/elaine/worker.rb

@@ -0,0 +1,25 @@
+module Elaine
+  class Worker
+    attr_reader :vertices, :active
+
+    def initialize(graph = [])
+      raise 'empty worker graph' if graph.empty?
+      @vertices = graph
+      @active   = graph.size
+    end
+
+    def superstep
+      Thread.new do
+        @vertices.each do |v|
+          v.messages = PostOffice.instance.read(v.id)
+          v.active! if v.messages.size > 0
+        end
+
+        active = @vertices.select {|v| v.active?}
+        active.each {|v| v.step}
+
+        @active = active.select {|v| v.active?}.size
+      end
+    end
+  end
+end

data/spec/coordinator_spec.rb

@@ -0,0 +1,95 @@
+require 'helper'
+
+describe Coordinator do
+  it 'should not allow empty graphs' do
+    lambda { Coordinator.new([]) }.should raise_error
+  end
+
+  let(:graph) do
+    [
+      AddVertex.new(:igvita,     1,  :wikipedia),
+      AddVertex.new(:wikipedia,  2,  :google),
+      AddVertex.new(:google,     1,  :wikipedia)
+    ]
+  end
+
+  it 'should partition graphs with variable worker sizes' do
+    c = Coordinator.new(graph)
+    c.workers.size.should == 1
+
+    c = Coordinator.new(graph, partitions: 2)
+    c.workers.size.should == 2
+  end
+
+  it 'should schedule workers to run until there are no active vertices' do
+    c = Coordinator.new(graph)
+    c.run
+
+    c.workers.each do |w|
+      w.vertices.each do |v|
+        v.value.should == 5
+      end
+    end
+  end
+
+  context 'PageRank' do
+    class PageRankVertex < Vertex
+      def compute
+        if superstep >= 1
+          sum = messages.inject(0) {|total,msg| total += msg; total }
+          @value = (0.15 / 3) + 0.85 * sum
+        end
+
+        if superstep < 30
+          deliver_to_all_neighbors(@value / neighbors.size)
+        else
+          halt
+        end
+      end
+    end
+
+
+    it 'should calculate PageRank of a circular graph' do
+      graph = [
+        #                   name     value  out-edges
+        PageRankVertex.new(:igvita,     1,  :wikipedia),
+        PageRankVertex.new(:wikipedia,  1,  :google),
+        PageRankVertex.new(:google,     1,  :igvita)
+      ]
+
+      c = Coordinator.new(graph)
+      c.run
+
+      c.workers.each do |w|
+        w.vertices.each do |v|
+          (v.value * 100).to_i.should == 33
+        end
+      end
+    end
+
+    it 'should calculate PageRank of arbitrary graph' do
+      graph = [
+        # page 1 -> page 1, page 2  (0.18)
+        # page 2 -> page 1, page 3  (0.13)
+        # page 3 -> page 3          (0.69)
+
+        #                   name     value  out-edges
+        PageRankVertex.new(:igvita,     1,  :igvita, :wikipedia),
+        PageRankVertex.new(:wikipedia,  1,  :igvita, :google),
+        PageRankVertex.new(:google,     1,  :google)
+      ]
+
+      c = Coordinator.new(graph)
+      c.run
+
+      c.workers.each do |w|
+        (w.vertices.find {|v| v.id == :igvita }.value * 100).ceil.to_i.should == 19
+        (w.vertices.find {|v| v.id == :wikipedia }.value * 100).ceil.to_i.should == 13
+        (w.vertices.find {|v| v.id == :google }.value * 100).to_i.should == 68
+      end
+    end
+  end
+
+  it 'should parition nodes by hashing the node id'
+  it 'should allow scheduling multiple partitions to a single worker'
+end

data/spec/distributed_coordinator_spec.rb

@@ -0,0 +1,143 @@
+require 'distributed_helper'
+
+describe Elaine::Distributed::Coordinator do
+  # it 'should not allow empty graphs' do
+  #   lambda { Coordinator.new([]) }.should raise_error
+  # end
+
+  before(:each) do
+
+    TestCoordinator.start
+    TestCoordinator.wait_until_ready
+    TestWorker1.start
+    TestWorker1.wait_until_ready
+    TestWorker2.start
+    TestWorker2.wait_until_ready
+  end
+
+  after(:each) do
+    TestWorker1.stop
+    TestWorker2.stop
+    TestCoordinator.stop
+  end
+
+  let(:graph) do
+    [
+      {
+        klazz: DistributedAddVertex,
+        id: :igvita,
+        value: 1,
+        outedges: [:wikipedia]
+      },
+      {
+        klazz: DistributedAddVertex,
+        id: :wikipedia,
+        value: 2,
+        outedges: [:google]
+      },
+      {
+        klazz: DistributedAddVertex,
+        id: :google,
+        value: 1,
+        outedges: [:wikipedia]
+      }
+    ]
+  end
+
+
+ 
+
+  it "should schedule workers to run intil there are no active vertices" do
+    
+    DCell::Node["test.elaine.coordinator"][:coordinator].graph = graph
+    DCell::Node["test.elaine.coordinator"][:coordinator].partition
+    DCell::Node["test.elaine.coordinator"][:coordinator].run_job
+    values = DCell::Node["test.elaine.coordinator"][:coordinator].vertex_values
+    values.each do |v|
+      v[:value].should == 5
+    end
+
+  end
+
+  it "should calculate PageRank of a circular graph" do
+    g =[
+      {
+        klazz: DistributedPageRankVertex,
+        id: :igvita,
+        value: 1,
+        outedges: [:wikipedia]
+      },
+      {
+        klazz: DistributedPageRankVertex,
+        id: :wikipedia,
+        value: 1,
+        outedges: [:google]
+      },
+      {
+        klazz: DistributedPageRankVertex,
+        id: :google,
+        value: 1,
+        outedges: [:igvita]
+      }
+    ]
+
+    DCell::Node["test.elaine.coordinator"][:coordinator].graph = g
+    DCell::Node["test.elaine.coordinator"][:coordinator].partition
+    DCell::Node["test.elaine.coordinator"][:coordinator].run_job
+
+    values = DCell::Node["test.elaine.coordinator"][:coordinator].vertex_values
+    values.each do |v|
+      (v[:value] * 100).to_i.should == 33
+    end
+  end
+  
+
+  it "should calculate PageRank of an arbitrary graph" do
+    g = [
+      # page 1 -> page 1, page 2  (0.18)
+      # page 2 -> page 1, page 3  (0.13)
+      # page 3 -> page 3          (0.69)
+
+      #                   name     value  out-edges
+      {
+        klazz: DistributedPageRankVertex,
+        id: :igvita,
+        value: 1,
+        outedges: [:igvita, :wikipedia]
+      },
+      {
+        klazz: DistributedPageRankVertex,
+        id: :wikipedia,
+        value: 1,
+        outedges: [:igvita, :google]
+      },
+      {
+        klazz: DistributedPageRankVertex,
+        id: :google,
+        value: 1,
+        outedges: [:google]
+      }
+    ]
+    DCell::Node["test.elaine.coordinator"][:coordinator].graph = g
+    DCell::Node["test.elaine.coordinator"][:coordinator].partition
+    DCell::Node["test.elaine.coordinator"][:coordinator].run_job
+
+    vertex_values = DCell::Node["test.elaine.coordinator"][:coordinator].vertex_values
+    vertex_values.each do |v|
+      if v[:id] == :igvita
+        (v[:value] * 100).ceil.to_i.should == 19
+      elsif v[:id] == :wikipedia
+        (v[:value] * 100).ceil.to_i.should == 13
+      elsif v[:id] == :google
+        (v[:value] * 100).to_i.should == 68
+      else
+        fail "Unexpected node id: #{v[:id]}"
+      end
+    end
+
+    
+  end
+
+  it 'should parition nodes by hashing the node id'
+  it 'should allow scheduling multiple partitions to a single worker'
+end

data/spec/distributed_helper.rb

@@ -0,0 +1,196 @@
+require 'elaine'
+require 'elaine/distributed'
+require 'dcell'
+
+# DCell.start
+
+class DistributedAddVertex < Elaine::Distributed::Vertex
+  def compute
+    @value += 1
+    halt if @value >= 5
+  end
+end
+
+class DistributedPageRankVertex < Elaine::Distributed::Vertex
+  def compute
+    if superstep >= 1
+      sum = messages.inject(0) {|total,msg| total += msg; total }
+      @value = (0.15 / 3) + 0.85 * sum
+    end
+
+    if superstep < 30
+      deliver_to_all_neighbors(@value / neighbors.size)
+    else
+      halt
+    end
+  end
+end
+
+
+module TestCoordinator
+  PORT = 8090
+  def self.start
+    @@pid = Process.spawn Gem.ruby, File.expand_path("../test_coordinator_node.rb", __FILE__)
+    puts "Coordinator pid: #{@@pid}"
+    unless @@pid
+      STDERR.print "ERROR: Couldn't start test coordinator node"
+      exit 1
+    end
+  end
+
+  def self.wait_until_ready
+    STDERR.print "Waiting for test coordinator node to start up..."
+
+    socket = nil
+    30.times do
+      begin
+        socket = TCPSocket.open("127.0.0.1", PORT)
+        break if socket
+      rescue Errno::ECONNREFUSED
+        STDERR.print "."
+        sleep 1
+      end
+    end
+
+    if socket
+      STDERR.puts " done!"
+      socket.close
+    else
+      STDERR.puts " FAILED!"
+      raise "couldn't connect to test node!"
+    end
+  end
+
+  def self.stop
+    puts "@@pid: #{@@pid}"
+    unless @@pid
+      STDERR.print "ERROR: Test coordinator node was never started!"
+      exit 1
+    end
+    Process.kill 9, @@pid
+  rescue Errno::ESRCH
+  ensure
+    Process.wait @@pid rescue nil
+  end
+end
+
+module TestWorker1
+  PORT = 8091
+  def self.start
+    @pid = Process.spawn Gem.ruby, File.expand_path("../test_worker_node1.rb", __FILE__)
+
+    unless @pid
+      STDERR.print "ERROR: Couldn't start test worker node 1"
+      exit 1
+    end
+  end
+
+  def self.wait_until_ready
+    STDERR.print "Waiting for test worker node 1 to start up..."
+
+    socket = nil
+    30.times do
+      begin
+        socket = TCPSocket.open("127.0.0.1", PORT)
+        break if socket
+      rescue Errno::ECONNREFUSED
+        STDERR.print "."
+        sleep 1
+      end
+    end
+
+    if socket
+      STDERR.puts " done!"
+      socket.close
+    else
+      STDERR.puts " FAILED!"
+      raise "couldn't connect to test node!"
+    end
+  end
+
+  def self.stop
+    unless @pid
+      STDERR.print "ERROR: Test worker node 1 was never started!"
+      exit 1
+    end
+    Process.kill 9, @pid
+  rescue Errno::ESRCH
+  ensure
+    Process.wait @pid rescue nil
+  end
+
+
+end
+
+
+module TestWorker2
+  PORT = 8092
+  def self.start
+    @pid = Process.spawn Gem.ruby, File.expand_path("../test_worker_node2.rb", __FILE__)
+
+    unless @pid
+      STDERR.print "ERROR: Couldn't start test worker node 2"
+      exit 1
+    end
+  end
+
+  def self.wait_until_ready
+    STDERR.print "Waiting for test worker node 2 to start up..."
+
+    socket = nil
+    30.times do
+      begin
+        socket = TCPSocket.open("127.0.0.1", PORT)
+        break if socket
+      rescue Errno::ECONNREFUSED
+        STDERR.print "."
+        sleep 1
+      end
+    end
+
+    if socket
+      STDERR.puts " done!"
+      socket.close
+    else
+      STDERR.puts " FAILED!"
+      raise "couldn't connect to test node!"
+    end
+  end
+
+  def self.stop
+    unless @pid
+      STDERR.print "ERROR: Test worker node 2 was never started!"
+      exit 1
+    end
+    Process.kill 9, @pid
+  rescue Errno::ESRCH
+  ensure
+    Process.wait @pid rescue nil
+  end
+
+
+end
+
+
+RSpec.configure do |config|
+  config.before(:suite) do
+    DCell.setup
+    DCell.run!    
+  end
+
+  # config.before(:all) do
+
+  #   TestCoordinator.start
+  #   TestCoordinator.wait_until_ready
+  #   TestWorker1.start
+  #   TestWorker1.wait_until_ready
+  #   TestWorker2.start
+  #   TestWorker2.wait_until_ready
+  # end
+
+  # config.after(:all) do
+  #   TestWorker1.stop
+  #   TestWorker2.stop
+  #   TestCoordinator.stop
+  # end
+end