@@ -7,8 +7,14 @@ include Gisting::MapReduce # TODO: Grab this from ARGV def args args = [] - # args = ["/Users/mchung/Public/datasets/aoldb_dev-aa", "/Users/mchung/Public/datasets/aoldb_dev-ab"] - args += ["/Users/mchung/Public/datasets/sample1.data", "/Users/mchung/Public/datasets/sample2.data"] + args += [ + "/Volumes/gisting/datasets/aol-data/user-ct-test-collection-01.txt", + "/Volumes/gisting/datasets/aol-data/user-ct-test-collection-02.txt", + "/Volumes/gisting/datasets/aol-data/user-ct-test-collection-03.txt", + "/Volumes/gisting/datasets/aol-data/user-ct-test-collection-04.txt", + "/Volumes/gisting/datasets/aol-data/user-ct-test-collection-05.txt", + ] + # args += ["/Users/mchung/Public/datasets/sample1.data", "/Users/mchung/Public/datasets/sample2.data"] # args += ["/Users/mchung/Public/datasets/aoldb_dev.txt"] end @@ -25,7 +31,7 @@ if __FILE__ == $0 end end output = spec.output - output.filebase = "/Users/mchung/Public/datasets/output" + output.filebase = "/Volumes/gisting/datasets/output" output.num_tasks = 2 output.reduce do |reduce_input| count = 0 examples/term_count.rb @@ -14,7 +14,7 @@ module Gisting def map_client_sent!(file_pattern) @lock.synchronize do @job_progress.start_map_job(file_pattern) - puts "MapServer done with #{file_pattern}" + puts "MapServer processing #{file_pattern}" puts "All Map Jobs Distributed" if @job_progress.all_map_jobs_in_progress? end end @@ -49,7 +49,7 @@ module Gisting def reduce_client_sent!(file_pattern) @lock.synchronize do @job_progress.start_reduce_job(file_pattern) - puts "ReduceServer done with #{file_pattern}" + puts "ReduceServer processing #{file_pattern}" puts "All Reduce Jobs Distributed" if @job_progress.all_reduce_jobs_in_progress? end end lib/gisting/job.rb 9c8964f79d9d83f5d1038aa7c0eadd837b1b6e3a Marc Chung mchung@gmail.com http://github.com/mchung/gisting/commit/a06f1ae8d77b04658ff37d2c9c6c2fdaeae67aa8 a06f1ae8d77b04658ff37d2c9c6c2fdaeae67aa8 2008-11-12T09:36:12-08:00 2008-11-12T09:36:12-08:00 Cleaning up puts statements dd8abd7ac4eee053765cd7ddeac056fffb587bcd Marc Chung mchung@gmail.com