gistfile1.rb

# last version had a bug:
#   instead of inserting 1 million docs,
#   it inserted 8000 docs then updated them 125 times

require 'rubygems'
require 'patron' 
require 'json'
 
COUCH = Patron::Session.new
COUCH.timeout = 120
COUCH.base_url = "http://localhost:5984"
COUCH.headers['User-Agent'] = 'ruby/mattetti' 
BATCH_SIZE = 8_000
RECORD_COUNT = 1_000_000
$COUCH_ID = 0
 
RECORD = {
   "component_version" => "X",
   "state" => "Analyze",
   "assignee" => "3002",
   "snapshot_date" => "20080701",
   "legacy_id" => "1318600",
   "component_name" => "Something Core",
   "priority" => "3"}
 
def bulk_of_docs
  records = []
  BATCH_SIZE.times do |n|
    records << RECORD.merge('_id' => ($COUCH_ID += 1).to_s)
  end
  records
end      

COUCH.put("/apple", {})

timer_start = Time.now 
(RECORD_COUNT / BATCH_SIZE).times do
  COUCH.post("/apple/_bulk_docs", {:docs => bulk_of_docs}.to_json)
end 
timer_end = Time.now 

COUCH.delete("/apple")

p "inserted #{RECORD_COUNT} documents in #{timer_end - timer_start}"