orphaned.rb

#!/usr/bin/ruby

# Utility for Puppet that outputs orphaned resources. Orphaned are
# resources that where at one point managed by Puppet, but they no
# longer are.
#


# We need to require Puppet because deserializing the report produces
# Puppet objects.
require 'optparse'
require 'yaml'
require 'puppet'

FILTERED_RESOURCES = ['Schedule[', 'File[/etc/puppet', 'File[/var/lib/puppet',
                      'File[/var/log/puppet', 'File[/var/run/puppet',
                      'configuration', 'Stage[main]', 'Class[Main]',
                      'Class[Settings]', 'Filebucket[puppet]']

# puppet config print lastrunreport
$report_path = "/var/lib/puppet/state/last_run_report.yaml"

# puppet config print lastrunfile
$summary_path = "/var/lib/puppet/state/last_run_summary.yaml"

# puppet config print statefile
$state_path = "/var/lib/puppet/state/state.yaml"

METHODS = %w[report timestamps]
$method = "report"

OptionParser.new do |opts|
  opts.banner = "Usage: orphaned.rb [options]"

  opts.on('-m', '--method TYPE', "report or timestamps. Always used. Default: #{$method}", METHODS) { |m| $method = m }
  opts.on('-t', '--state PATH', "Path to state.yaml. Default: #{$state_path}") { |p| $state_path = p }
  opts.on('-r', '--report PATH', "Path to last_run_report.yaml. Used in method report. Default: #{$report_path}") { |p| $report_path = p }
  opts.on('-s', '--summary PATH', "Path to last_run_summary.yaml. Used in method timestamps. Default: #{$summary_path}") { |p| $summary_path = p }
  opts.on('-p', '--previous-report PATH', "Path to a previous last_run_report.yaml. If given, output has more detail.") { |p| $previous_report_path = p }

end.parse!

def parse_last_run(last_run_summary_data)
  total = last_run_summary_data['time']['total']
  seconds_last_run = last_run_summary_data['time']['last_run']
  Time.at(seconds_last_run - total - 2)
end

def get_last_run(last_run_summary_path)
  data = YAML.load_file(last_run_summary_path)
  parse_last_run(data)
end

def parse_state_resources(state_data)
  resources_data = state_data.select do |k, v|
    not k.to_s.start_with?(*FILTERED_RESOURCES)
  end

  if resources_data.is_a? Array
    resources_data.to_h
  else
    resources_data
  end
end

def get_state_resources(state_path)
  data = YAML.load_file(state_path)
  parse_state_resources(data)
end

def parse_old_state_resources(state_data, last_run)
  resources_data = parse_state_resources(state_data)

  unmanaged_resources = []
  resources_data.each do |k, v|
    if v[:checked] < last_run
      unmanaged_resources.push(k)
    end
  end
  
  return unmanaged_resources
end

def get_old_state_resources(state_path, last_run)
  data = YAML.load_file(state_path)
  parse_old_state_resources(data, last_run)
end

def parse_report_resources(report_data)
  resources_data = report_data.resource_statuses.select do |k, v|
    not k.to_s.start_with?(*FILTERED_RESOURCES)
  end

  if resources_data.is_a? Array
    resources_data.to_h
  else
    resources_data
  end
end

def get_report_resources(report_path)
  data = YAML.load_file(report_path)
  parse_report_resources(data)
end

def get_orphans_using_timestamps(state_path, summary_path)
  last_run = get_last_run(summary_path)
  get_old_state_resources(state_path, last_run)
end

def get_orphans_using_report(state_path, report_path)
  state_resources = get_state_resources(state_path)
  report_resources = get_report_resources(report_path)
  state_resources.keys - report_resources.keys
end

def print_detail(orphan_resources, previous_report_path)
  previous_resources = get_report_resources(previous_report_path)

  orphan_resources.each do |k|
    puts k
    if previous_resources.keys.include? k
      puts previous_resources[k].to_yaml
    end
  end

end

if $method == 'report'
  orphans = get_orphans_using_report($state_path, $report_path)
else
  orphans = get_orphans_using_timestamps($state_path, $summary_path)
end

if $previous_report_path.nil?
  puts orphans
else
  print_detail(orphans, $previous_report_path)
end