replay_es_index.rb

#!/usr/bin/env ruby

# Throwaway script to dump an index's contents to disk, delete the index, then
# re-upload it. Used for testing index templates in Logstash.
#
# Installing the yajl gem is highly recommended for fast JSON, but not required.

require 'rubygems'
require 'yajl/json_gem' rescue require 'json'
require 'elasticsearch'
require 'active_support/core_ext/array/grouping'

def index_file_name
  "#{@index_name}.json"
end

def get_index_data
  hits = @client.search(index: @index_name, size: 0)['hits']['total']
  puts "#{hits} documents in #{@index_name}"
  results = @client.search(index: @index_name, size: hits)
  data = results['hits']['hits'].map do |hit|
    { index: { _index: @index_name, _type: hit['_type'], _id: hit['_id'], data: hit['_source'] } }
  end

  File.open(index_file_name, 'w+') do |f|
    f.write(data.to_json)
  end
  puts "wrote #{index_file_name} for later use"
  data
end

@index_name = ARGV[0]

raise "Syntax: replay_es_index.rb INDEX_NAME" if @index_name.nil?

@client = Elasticsearch::Client.new

index_data = if File.exist? index_file_name
  JSON.parse(File.read(index_file_name))
else
  get_index_data
end

puts "Deleting #{@index_name}  if it exists"
@client.indices.delete index: @index_name rescue nil

puts "Bulk inserting #{@index_name} (in groups of 10,000)"
groups = index_data.in_groups_of 10_000, false

groups.each_with_index do |group, idx|
  @client.bulk body: group
  puts "inserted group #{idx + 1} of #{groups.count}"
end