cookbooks/planet/templates/default/planet-file-cleanup.erb

   1 #!/usr/bin/ruby
   2
   3 require 'date'
   4 require 'optparse'
   5
   6 # always keep the last 4 weeks
   7 ALWAYS_KEEP_DAYS = 4 * 7
   8
   9 # otherwise, bucket by month and keep the earliest in the bucket
  10 def bucket(date)
  11   Date.new(date.year, date.month, 1)
  12 end
  13
  14 Candidate = Struct.new(:filename, :date)
  15
  16 def list_files(glob, date_pattern)
  17   # find all candidates for deletion
  18   real_files = Dir.glob(glob).select do |file|
  19     File.file?(file) && !File.symlink?(file)
  20   end
  21
  22   real_files.map do |file|
  23     # extract the date
  24     m = date_pattern.match(file)
  25     raise "Unable to extract date string from #{file.inspect}" if m.nil?
  26     d = Date.strptime(m[1], "%y%m%d")
  27     Candidate.new(file, d)
  28   end
  29 end
  30
  31 def deletion_candidates(today, candidates)
  32   candidate_buckets = Hash.new
  33
  34   candidates.each do |c|
  35     next if today - c.date < ALWAYS_KEEP_DAYS
  36     b = bucket(c.date)
  37
  38     candidate_buckets[b] = Array.new unless candidate_buckets.has_key?(b)
  39     candidate_buckets[b] << c
  40   end
  41
  42   # delete all but the earliest in each bucket
  43   candidate_buckets.collect_concat do |bucket, contents|
  44     contents.sort_by {|c| c.date}[1..-1]
  45   end
  46 end
  47
  48 def deletions(glob, date_pattern, today, expansions)
  49   candidates = list_files(glob, date_pattern)
  50   to_delete = deletion_candidates(today, candidates)
  51
  52   expanded = to_delete.collect_concat do |candidate|
  53     dir = File.dirname(candidate.filename)
  54     expansions.map do |e|
  55       exp = candidate.date.strftime(e)
  56       "#{dir}/#{exp}"
  57     end
  58   end
  59
  60   expanded.select {|e| File.exist?(e)}
  61 end
  62
  63 dry_run = false
  64 debug = false
  65
  66 OptionParser.new do |opt|
  67   opt.on('--dry-run') { dry_run = true }
  68   opt.on('--debug') { debug = true }
  69 end.parse!
  70
  71 xml_directory = "<%= node[:planet][:dump][:xml_directory] %>"
  72 xml_history_directory = "<%= node[:planet][:dump][:xml_history_directory] %>"
  73 pbf_directory = "<%= node[:planet][:dump][:pbf_directory] %>"
  74 pbf_history_directory = "<%= node[:planet][:dump][:pbf_history_directory] %>"
  75
  76 today = Date.today
  77 to_delete = Array.new
  78
  79 to_delete += deletions(
  80   "#{xml_directory}/20??/planet-??????.osm.bz2",
  81   /planet-([0-9]{6}).osm.bz2/,
  82   today,
  83   ["changesets-%y%m%d.osm.bz2",
  84    "changesets-%y%m%d.osm.bz2.md5",
  85    "discussions-%y%m%d.osm.bz2",
  86    "discussions-%y%m%d.osm.bz2.md5",
  87    "planet-%y%m%d.osm.bz2",
  88    "planet-%y%m%d.osm.bz2.md5"])
  89
  90 to_delete += deletions(
  91   "#{xml_history_directory}/20??/history-??????.osm.bz2",
  92   /history-([0-9]{6}).osm.bz2/,
  93   today,
  94   ["history-%y%m%d.osm.bz2",
  95    "history-%y%m%d.osm.bz2.md5"])
  96
  97 to_delete += deletions(
  98   "#{pbf_directory}/planet-??????.osm.pbf",
  99   /planet-([0-9]{6}).osm.pbf/,
 100   today,
 101   ["planet-%y%m%d.osm.pbf",
 102    "planet-%y%m%d.osm.pbf.md5"])
 103
 104 to_delete += deletions(
 105   "#{pbf_history_directory}/history-??????.osm.pbf",
 106   /history-([0-9]{6}).osm.pbf/,
 107   today,
 108   ["history-%y%m%d.osm.pbf",
 109    "history-%y%m%d.osm.pbf.md5"])
 110
 111 total_size = 0
 112 num_deleted = 0
 113 cmd = dry_run ? "Would delete" : "Deleted"
 114 to_delete.each do |file|
 115   s = File.stat(file)
 116   File.delete(file) unless dry_run
 117   puts "#{cmd} #{file.inspect}, #{s.size / 1000000} MB" if debug
 118   total_size += s.size
 119   num_deleted += 1
 120 end
 121 if debug
 122   puts "#{cmd} files of total size #{total_size / 1000000000.0} GB"
 123   puts "#{cmd} #{num_deleted} files"
 124 end