[Chef] Get Bookshelf Usage by-org

bookshelf-du.rb

# Retrieve the size of Chef cookbooks used by different organizations on the Chef server it runs on
# Should run as root normally since we use su to hop across to another user

full_org_list=''
full_file_list=''
orgs={}

# Getting orgs can take a while, let's hold onto the results for repeat invocations
if not File.exist?('.orgs-cached')
# NOTE: this produces 2 header lines and a footer line that has the rowcount
  full_org_list=`su -l opscode-pgsql -c \"psql -d opscode_chef -c 'SELECT * from orgs'\"| tail -n +3 | sed -e 's/\s|\s/|/g'`
  File.open(".orgs-cached","w") do |f|
    f.puts full_org_list
  end
else
  full_org_list=IO.readlines('.orgs-cached')
end

# Getting a large bookshelf directory takes a significant period of time
if not File.exist?('.files-cached')
  full_file_list=`find /var/opt/opscode/drbd/data/bookshelf/ -mindepth 2 -type f -printf \"%p|%k\n\" | sort -k 12 -t /`
  File.open(".files-cached", "w") do |f|
    f.puts full_file_list
  end
else
  full_file_list=IO.readlines('.files-cached')
end

full_org_list.each do |org_line|
  org_cols = org_line.split('|')
  next if org_cols.length != 8
  org_cols[0].strip!
# column 2 is name, column 0 is the id
  orgs[org_cols[0]] = org_cols[2].strip
end

org_usage={}
full_file_list.each do |file_line|
  /organization\-(\w+).*\|(\d+)$/ =~ file_line
  org_id=$1
  next if not org_id
  accum_sz = org_usage[org_id] || 0
  org_usage[org_id] = accum_sz.to_i + $2.to_i
end

total_use = 0
org_usage.each do |k,v|
  org_label = orgs[k] || k
  puts "%-40s Usage: %10d KB" % [org_label, v]
  total_use += v
end
puts "------------\n%-40s Total: %10d KB" % ["", total_use]