Is it possible to automatically dump test/development data from a production rails app? Here's a rake task (./lib/tasks/dump.rake) that starts trying to do so.

dump.rake

require 'rubygems'
require 'csv'
require 'pry'

desc "Dump a representative set of test/dev data"
task :dump => :environment do

  #Load models so they are available for the ActiveRecord::Base.descendants call
  Dir.foreach("#{Rails.root}/app/models") do |f|
    require f if f =~ /.*\.rb/ 
  end
  
  #Build a hash of table names to model names
  models = ActiveRecord::Base.descendants.reduce({}) do |hash, d| 
    hash[d.table_name] = d.name
    hash 
  end
  
  #Grab all tables according to the DB (might not match our models exactly)
  tables = ActiveRecord::Base.connection.tables
  
  #Remember which tables don't have models
  no_model_tables = tables.select do |t|
    ! models.include? t 
  end
  
  #Dump table names, model names, and record counts to a CSV for perusal
  CSV.open('table_counts.csv', 'wb') do |csv|
    tables.each do |t|
      csv << [t, models[t], ActiveRecord::Base.connection.execute("select count(*) from #{t}").first.first]
    end
  end
  
  #load CSV for binding.pry
  table_model_count = CSV.read('table_counts.csv')
  
  #Open a ruby shell to play with data
  binding.pry
end