find_dupe_uuids.rb

#!/opt/delivery/embedded/bin/ruby

require 'json'
require 'pp'

days = ARGV.first
days ||= 3

STDERR.puts "Fetching #{days} days of compliance indices"
all_indices = JSON.parse(`curl -s -k -XGET 'localhost:9200/_cat/indices?format=json'`)
indices = all_indices.select{ |i| i['index'].include?('compliance-2')  }.map{ |i| i['index'] }.sort.reverse[0...days.to_i]

STDERR.puts "Found the following indices: #{indices.inspect}"

STDERR.puts "Fetching data from ES"

data = `curl -s -k -XGET 'localhost:9200/#{indices.join(',')}/_search?size=10000'`

json = JSON.parse(data)

STDERR.puts "Parsing ES results"
node_list = json['hits']['hits'].inject({}) do |collector, node|
  source = node['_source']
  uuid = source['node_uuid']

  if !uuid.nil?
	  collector[uuid] ||= []
	  node_info = { name: source['node_name'] }

	  collector[uuid] <<  node_info unless collector[uuid].include?(node_info)
  end

  collector
end

node_list.reject! { |id,nodes| nodes.count <= 1 }

if node_list.keys.count > 0
	STDERR.puts "Here are the nodes with duplicate data collector UUIDs"
	node_list.values.flatten.each do |node|
	  puts node[:name]
        end
else
	STDERR.puts "No nodes with duplicate data collector uuids were where found"
end