A simple script to find the fastest mirrors from ubuntu's mirrors.txt

mirror_filter.rb

#!/usr/bin/env ruby
# encoding: utf-8

require 'rubygems'
require 'net/http'
require 'net/ping/tcp'

# This script grabs the master mirror list file from mirrors.ubuntu.com
# And then attempts to find the fastest subset of them and writes it out to a file
# This is intended to be used with the apt mirrors method in order to get the fastest
# updates to our vagrant VM 
# Reference: http://mvogt.wordpress.com/2011/03/21/the-apt-mirror-method/

# Change this to output to a different file
OUTPUT_FILE='/var/www/nginx-default/mirrors.txt'

mirrorlist_host = 'mirrors.ubuntu.com'
min_num_servers=5 # Minimum number of servers to require in server list
timeout=0.05 # start with 50ms limit
max_tries=3  # Number of times to try to get min_num_servers servers with <timeout ping
timeout_delta=0.01 # timeout increment to use for retry
uri_list = []

# The technique used here is:
# 1. loop through each mirror
# 2. Trim list by performing TCP ping on port 80 for each (with a timeout limit)
# 3. If we end up with less than min_num_servers,  and tries is < max_tries then retry with same timeout
#    Else increase timeout by timeout_delta and retry

# Define ShortServerListException so we can catch it and retry
class ShortServerListException < Exception
end

Net::HTTP.start(mirrorlist_host) do |http|
  resp = http.get("/mirrors.txt")

  resp.body.each_line do |line|
    uri_list.push( URI(line) )
  end
end

puts "Total Mirrors Found: #{uri_list.length}"

all_uris = uri_list.clone

File.open(OUTPUT_FILE, 'wt') do |f|
  try_count=0
  # Remove hosts with > timeout ping (in ms)
  begin
    uri_list.delete_if do |uri|
      # puts "LENGTH: #{uri_list.length}"
      # puts uri_list
      if Net::Ping::TCP.new('mirror.vcu.edu',80, timeout).ping
        printf "%40s\t%s\n", uri.host, "\033[1;32m✓\033[0m"
        f.write(uri.to_s + "\n")
          false
      else
        printf "%40s\t%s\n", uri.host, "\033[1;31m✗\033[0m"
        # puts "Removing index: #{uri_list.find_index(uri)}"
        true
      end
    end


    try_count += 1
    # puts "FINAL LIST:"
    # puts uri_list
    if uri_list.length.zero? || uri_list.length < min_num_servers
      f.truncate(0)
      msg = "Not enough mirror URIs failed to pass ping test!\nA minimum of #{min_num_servers} passing servers is required"
      puts msg
      raise ShortServerListException, msg
    end 
  rescue ShortServerListException
    # restart with full list
    uri_list = all_uris.clone
    puts "Retrying with full url list"
    puts "TRIES: #{try_count}"
    puts "LEN  : #{uri_list.length}"
    puts "ZERO?: #{uri_list.length.zero?}"
    if try_count < max_tries
      retry
    else
      timeout += timeout_delta
      puts "Increasing timeout: #{timeout}"
      retry
    end
  end

end