Update log analysis script to use Maxmind GeoIP data

[dns.git] / bin / mkgeo

diff --git a/bin/mkgeo b/bin/mkgeo
index fd0459307a2b2c7a13ee2b9533d522e9180fece1..4bd4db2f8a9d3ea16c19409f56d45939a57ddfce 100755 (executable)
--- a/bin/mkgeo
+++ b/bin/mkgeo
@@ -7,38 +7,76 @@ use IO::File;
 use Math::Trig qw(deg2rad pip2 great_circle_distance);
 use JSON::XS;
 use LWP::UserAgent;
 use Math::Trig qw(deg2rad pip2 great_circle_distance);
 use JSON::XS;
 use LWP::UserAgent;

-use XML::TreeBuilder;

 use YAML;
 
 use YAML;
 

-my $source = shift @ARGV;
+my $originfile = shift @ARGV;
+my $clusterfile = shift @ARGV;

 my $zone = shift @ARGV;
 my $zone = shift @ARGV;

-my $servers = YAML::LoadFile("src/${source}");
+my $targetoriginfile = shift @ARGV;
+my $origins = YAML::LoadFile($originfile);
+my $clusters = YAML::LoadFile($clusterfile);
+my @servers;

 
 

-# Initialise server details
-while (my($name,$server) = each %$servers)
+# Initialise cluster details
+while (my($name,$cluster) = each %$clusters)

 {
 {

-    $server->{name} = $name;
-    $server->{bandwidth} = $server->{bandwidth} * 1024 * 1024;
-
-    if ($ENV{PINGDOM_USERNAME} && $ENV{PINGDOM_PASSWORD})
+    if ($cluster->{servers})

     {
     {

-        $server->{status} = "down";
+        $cluster->{bandwidth} = 0;
+
+        foreach my $server (@{$cluster->{servers}})
+        {
+            $server->{cluster} = $cluster;
+            $cluster->{bandwidth} = $cluster->{bandwidth} + $server->{bandwidth};
+
+            push @servers, $server;
+        }

     }
     else
     {
     }
     else
     {

-        $server->{status} = "up";
+        my $server = {
+            cluster => $cluster,
+            pingdom => $cluster->{pingdom},
+            bandwidth => $cluster->{bandwidth},
+            ipv4 => $cluster->{ipv4},
+            ipv6 => $cluster->{ipv6}
+        };
+
+        $cluster->{servers} = [ $server ];
+
+        push @servers, $server;

     }
     }

+
+    $cluster->{name} = $name;
+    $cluster->{status} = "down";
+}
+
+# Initialise server details
+foreach my $server (@servers)
+{
+    $server->{status} = "up";

 }
 
 # If pingdom support is enabled then check which servers are up
 if ($ENV{PINGDOM_USERNAME} && $ENV{PINGDOM_PASSWORD})
 {
     my $ua = LWP::UserAgent->new;
 }
 
 # If pingdom support is enabled then check which servers are up
 if ($ENV{PINGDOM_USERNAME} && $ENV{PINGDOM_PASSWORD})
 {
     my $ua = LWP::UserAgent->new;

+    my $cache;

 
 

+    $ua->timeout(5);

     $ua->default_header("App-Key", "2cohi62u5haxvqmypk3ljqqrze1jufrh");
     $ua->credentials("api.pingdom.com:443", "Pingdom API", $ENV{PINGDOM_USERNAME}, $ENV{PINGDOM_PASSWORD});
 
     $ua->default_header("App-Key", "2cohi62u5haxvqmypk3ljqqrze1jufrh");
     $ua->credentials("api.pingdom.com:443", "Pingdom API", $ENV{PINGDOM_USERNAME}, $ENV{PINGDOM_PASSWORD});
 

-    foreach my $server (values %$servers)
+    if (-f "pingdom.yml")
+    {
+        $cache = YAML::LoadFile("pingdom.yml"); 
+    }
+    else
+    {
+        $cache = {};
+    }
+
+    foreach my $server (@servers)

     {
         if (my $checkid = $server->{pingdom})
         {
     {
         if (my $checkid = $server->{pingdom})
         {


@@ -49,85 +87,98 @@ if ($ENV{PINGDOM_USERNAME} && $ENV{PINGDOM_PASSWORD})
                 my $check = decode_json($response->content);
 
                 $server->{status} = $check->{check}->{status};
                 my $check = decode_json($response->content);
 
                 $server->{status} = $check->{check}->{status};

+                $cache->{$server->{pingdom}} = $check->{check}->{status};
+            }
+            else
+            {
+                $server->{status} = $cache->{$server->{pingdom}} || "down";

             }
         }
             }
         }

+        else
+        {
+            $server->{status} = "down";
+        }

     }
     }

-}

 
 

-my %countries = ();
-my @mappings = ();
-
-# Create a parser for the country database
-my $countries = XML::TreeBuilder->new;
+    YAML::DumpFile("pingdom.yml", $cache);
+}

 
 

-# Parse the country database
-$countries->parsefile("lib/countries.xml");
+# Mark a cluster as up if any servers are up
+foreach my $server (@servers)
+{
+    if ($server->{status} eq "up")
+    {
+        $server->{cluster}->{status} = "up";
+    }
+    else
+    {
+        $server->{cluster}->{bandwidth} = $server->{cluster}->{bandwidth} - $server->{bandwidth};
+    }
+}

 
 

-# Load the per-country bandwidth details
-my $bandwidth = YAML::LoadFile("bandwidth/${source}.yml");
+# Create target origins object
+my $targetorigins = {};

 
 

-# Fill in country table and work out which servers each can use
-foreach my $country ($countries->look_down("_tag" => "country"))
+# Initialise cluster details
+while (my($name,$cluster) = each %$clusters)

 {
 {

-    my $code = $country->look_down("_tag" => "countryCode")->as_text;
-    my $name = $country->look_down("_tag" => "countryName")->as_text;
-    my $population = $country->look_down("_tag" => "population")->as_text;
-    my $bandwidth = $bandwidth->{$code} || 0;
-    my $continent = $country->look_down("_tag" => "continent")->as_text;
-    my $west = $country->look_down("_tag" => "west")->as_text;
-    my $north = $country->look_down("_tag" => "north")->as_text;
-    my $east = $country->look_down("_tag" => "east")->as_text;
-    my $south = $country->look_down("_tag" => "south")->as_text;
-    my $lat = centre_lat( $south, $north );
-    my $lon = centre_lon( $west, $east );
-
-    $countries{$code} = {
-        code => $code, name => $name, continent => $continent,
-        bandwidth => $bandwidth, lat => $lat, lon => $lon
+    $cluster->{bandwidth_limit} = $cluster->{bandwidth} * 1024 * 1024;
+    $cluster->{bandwidth_used} = 0;
+
+    $targetorigins->{$cluster->{name}} = {
+        code => $cluster->{name},
+        name => $cluster->{name},
+        lat => $cluster->{lat},
+        lon => $cluster->{lon},
+        bandwidth => 0

     };
     };

+}
+
+my @mappings = ();

 
 

-    foreach my $server (values %$servers)
+# Scan origins and work out which clusters each can use
+foreach my $origin (values %$origins)
+{
+    foreach my $cluster (values %$clusters)

     {
     {

-        my $match = match_country($server, $code, $continent);
+        my $match = match_origin($cluster, $origin);

 
 

-        if ($server->{status} eq "up" && $match ne "denied")
+        if ($cluster->{status} eq "up" && $match ne "denied")

         {
             my $priority = $match eq "preferred" ? 20 : 10;
         {
             my $priority = $match eq "preferred" ? 20 : 10;

-            my $distance = distance($lat, $lon, $server->{lat}, $server->{lon});
+            my $distance = distance($origin->{lat}, $origin->{lon}, $cluster->{lat}, $cluster->{lon});

 
             push @mappings, {
 
             push @mappings, {

-                country => $countries{$code}, server => $server,
+                origin => $origin, cluster => $cluster,

                 priority => $priority, distance => $distance
             };
         }
     }
 }
 
                 priority => $priority, distance => $distance
             };
         }
     }
 }
 

-# Discard the parsed country database
-$countries->delete;
+# Allocate each country to a cluster
+allocate_clusters(@mappings);

 
 

-# Loop over the mappings, trying to assign each country to the
-# nearest server, but subject to the bandwidth limits;
-foreach my $mapping (sort {  $b->{priority} <=> $a->{priority} || $a->{distance} <=> $b->{distance} } @mappings)
+# If we failed to allocate every origin then loop, increasing
+# the bandwidth for each cluster by a little and retrying until
+# we manage to allocate everything
+while (grep { !exists($_->{cluster}) } values %$origins)

 {
 {

-    my $country = $mapping->{country};
-    my $server = $mapping->{server};
-
-    if ($country->{bandwidth} <= $server->{bandwidth} && !exists($country->{server}))
+    # Clear any existing mappings of countries to clusters
+    foreach my $origin (values %$origins)

     {
     {

-        $country->{server} = $server;
-        $server->{bandwidth} = $server->{bandwidth} - $country->{bandwidth};
+        delete $origin->{cluster};

     }
     }

-}

 
 

-# Loop over the mappings again, assigning anything that is left
-# as best we can, and allowing bandwidth limits to be exeeded
-foreach my $mapping (sort {  $b->{priority} <=> $a->{priority} || $a->{distance} <=> $b->{distance} } @mappings)
-{
-    my $country = $mapping->{country};
-    my $server = $mapping->{server};
+    # Reset bandwidth usage for clusters and increase limits by 10%
+    foreach my $cluster (values %$clusters)
+    {
+        $cluster->{bandwidth_used} = 0;
+        $cluster->{bandwidth_limit} = $cluster->{bandwidth_limit} * 1.1;
+    }

 
 

-    $country->{server} = $server unless exists($country->{server});
+    # Try the allocate again
+    allocate_clusters(@mappings);

 }
 
 # Create JSON collection object
 }
 
 # Create JSON collection object


@@ -138,13 +189,13 @@ my $zonefile = IO::File->new("> data/${zone}") || die "$!";
 my $jsonfile = IO::File->new("> json/${zone}.json") || die "$!";
 
 # Output details for each country
 my $jsonfile = IO::File->new("> json/${zone}.json") || die "$!";
 
 # Output details for each country

-foreach my $country (values %countries)
+foreach my $origin (values %$origins)

 {
 {

-    my $server = $country->{server};
-    my $clon = $country->{lon};
-    my $clat = $country->{lat};
-    my $slon = $server->{lon};
-    my $slat = $server->{lat};
+    my $cluster = $origin->{cluster};
+    my $clon = $origin->{lon};
+    my $clat = $origin->{lat};
+    my $slon = $cluster->{lon};
+    my $slat = $cluster->{lat};

 
     if ($clon > 0 && $slon < 0 && 360 + $slon - $clon < $clon - $slon)
     {
 
     if ($clon > 0 && $slon < 0 && 360 + $slon - $clon < $clon - $slon)
     {


@@ -155,7 +206,8 @@ foreach my $country (values %countries)
         $clon = $clon + 360;
     }
 
         $clon = $clon + 360;
     }
 

-    $zonefile->print("C\L$country->{code}\E.${zone}:$server->{name}.${zone}:600\n");
+    $zonefile->print("# $origin->{name}\n");
+    $zonefile->print("C\L$origin->{code}\E.${zone}:$cluster->{name}.${zone}:600\n");

 
     push @json, {
         type => "Feature",
 
     push @json, {
         type => "Feature",


@@ -164,15 +216,38 @@ foreach my $country (values %countries)
             coordinates => [ [ $clon, $clat ], [ $slon, $slat ] ]
         },
         properties => {
             coordinates => [ [ $clon, $clat ], [ $slon, $slat ] ]
         },
         properties => {

-            country => $country->{name}
+            origin => $origin->{name},
+            server => $cluster->{name},
+            colour => $cluster->{colour}

         }
     };
         }
     };

+
+    $targetorigins->{$cluster->{name}}->{bandwidth} += $origin->{bandwidth};

 }
 
 }
 

+# Header for default records
+$zonefile->print("# Unknown origins\n");
+

 # Output default records for IPs that can't be mapped to a country
 # Output default records for IPs that can't be mapped to a country

-foreach my $server (grep { $servers->{$_}->{default} } keys %$servers)
+while (my($name,$cluster) = each %$clusters)

 {
 {

-    $zonefile->print("Cxx.${zone}:${server}.${zone}:600\n");
+    if (my $default = $cluster->{default})
+    {
+        output_server($zonefile, "${default}.${zone}", $cluster);
+    }
+    elsif (exists($cluster->{default}))
+    {
+        output_server($zonefile, "${zone}", $cluster);
+    }
+}
+
+# Header for underlying servers
+$zonefile->print("# Servers\n");
+
+# Output A records for each cluster
+while (my($name,$cluster) = each %$clusters)
+{
+    output_server($zonefile, "${name}.${zone}", $cluster);

 }
 
 # Output the GeoJSON text
 }
 
 # Output the GeoJSON text


@@ -182,89 +257,57 @@ $jsonfile->print(encode_json(\@json));
 $jsonfile->close();
 $zonefile->close();
 
 $jsonfile->close();
 $zonefile->close();
 

-exit 0;
-
-#
-# Find the centre value between two latitudes
-#
-sub centre_lat
-{
-    my $south = shift;
-    my $north = shift;
-
-    return ( $south + $north ) / 2;
-}
-
-#
-# Find the centre value between two longitudes
-#
-sub centre_lon
-{
-    my $west = shift;
-    my $east = shift;
-    my $lon;
-
-    if ($west < $east)
-    {
-        $lon = ( $west + $east ) / 2;
-    }
-    else
-    {
-        $lon = ( $west + $east + 360 ) / 2;
-    }
-
-    $lon = $lon - 360 if $lon > 180;
+# Output the target details in origin format if required
+YAML::DumpFile($targetoriginfile, $targetorigins) if $targetoriginfile;

 
 

-    return $lon
-}
+exit 0;

 
 #
 
 #

-# Match a country against a server
+# Match an origin against a cluster

 #
 #

-sub match_country
+sub match_origin

 {
 {

-    my $server = shift;
-    my $country = shift;
-    my $continent = shift;
+    my $cluster = shift;
+    my $origin = shift;

     my $match;
 
     my $match;
 

-    if ($server->{preferred} &&
-        $server->{preferred}->{countries} &&
-        grep { $_ eq $country } @{$server->{preferred}->{countries}})
+    if ($cluster->{preferred} &&
+        $cluster->{preferred}->{countries} &&
+        grep { $_ eq $origin->{country} } @{$cluster->{preferred}->{countries}})

     {
         $match = "preferred";
     }
     {
         $match = "preferred";
     }

-    elsif ($server->{preferred} &&
-           $server->{preferred}->{continents} &&
-           grep { $_ eq $continent } @{$server->{preferred}->{continents}})
+    elsif ($cluster->{allowed} &&
+           $cluster->{allowed}->{countries} &&
+           grep { $_ eq $origin->{country} } @{$cluster->{allowed}->{countries}})

     {
     {

-        $match = "preferred";
+        $match = "allowed";

     }
     }

-    elsif ($server->{allowed} &&
-           $server->{allowed}->{countries} &&
-           grep { $_ eq $country } @{$server->{allowed}->{countries}})
+    elsif ($cluster->{denied} &&
+           $cluster->{denied}->{countries} &&
+           grep { $_ eq $origin->{country} } @{$cluster->{denied}->{countries}})

     {
     {

-        $match = "allowed";
+        $match = "denied";

     }
     }

-    elsif ($server->{allowed} &&
-           $server->{allowed}->{continents} &&
-           grep { $_ eq $continent } @{$server->{allowed}->{continents}})
+    elsif ($cluster->{preferred} &&
+           $cluster->{preferred}->{continents} &&
+           grep { $_ eq $origin->{continent} } @{$cluster->{preferred}->{continents}})

     {
     {

-        $match = "allowed";
+        $match = "preferred";

     }
     }

-    elsif ($server->{denied} &&
-        $server->{denied}->{countries} &&
-        grep { $_ eq $country } @{$server->{preferred}->{countries}})
+    elsif ($cluster->{allowed} &&
+           $cluster->{allowed}->{continents} &&
+           grep { $_ eq $origin->{continent} } @{$cluster->{allowed}->{continents}})

     {
     {

-        $match = "denied";
+        $match = "allowed";

     }
     }

-    elsif ($server->{denied} &&
-           $server->{denied}->{continents} &&
-           grep { $_ eq $continent } @{$server->{preferred}->{continents}})
+    elsif ($cluster->{denied} &&
+           $cluster->{denied}->{continents} &&
+           grep { $_ eq $origin->{continent} } @{$cluster->{denied}->{continents}})

     {
         $match = "denied";
     }
     {
         $match = "denied";
     }

-    elsif ($server->{allowed})
+    elsif ($cluster->{allowed})

     {
         $match = "denied";
     }
     {
         $match = "denied";
     }


@@ -288,3 +331,88 @@ sub distance
 
     return great_circle_distance($lon1, pip2 - $lat1, $lon2, pip2 - $lat2);
 }
 
     return great_circle_distance($lon1, pip2 - $lat1, $lon2, pip2 - $lat2);
 }

+
+#
+# Allocate each origin to a cluster
+#
+sub allocate_clusters
+{
+    my @mappings = sort { compare_mappings($a, $b) } @_;
+
+    # Loop over the mappings, trying to assign each origin to the
+    # nearest cluster, but subject to the bandwidth limits
+    while (my $mapping = shift @mappings)
+    {
+        my @group;
+
+        push @group, $mapping;
+
+        while (@mappings && compare_mappings($mapping, $mappings[0]) == 0)
+        {
+            push @group, shift @mappings;
+        }
+
+        for my $mapping (sort compare_bandwidth @group)
+        {
+            my $origin = $mapping->{origin};
+            my $cluster = $mapping->{cluster};
+
+            if (!exists($origin->{cluster}) &&
+                $cluster->{bandwidth_used} + $origin->{bandwidth} <= $cluster->{bandwidth_limit})
+            {
+                $origin->{cluster} = $cluster;
+                $cluster->{bandwidth_used} = $cluster->{bandwidth_used} + $origin->{bandwidth};
+            }
+        }
+    }
+
+    return;
+}
+
+#
+# Compare two mappings to decide which to use
+#
+sub compare_mappings
+{
+    my $a = shift;
+    my $b = shift;
+
+    return $b->{priority} <=> $a->{priority} ||
+           $a->{distance} <=> $b->{distance};
+}
+
+#
+# Compare two mappings to decide which to try first
+#
+sub compare_bandwidth
+{
+    my $a_used = ( $a->{cluster}->{bandwidth_used} * 100.0 ) / ( $a->{cluster}->{bandwidth_limit} * 1.0 );
+    my $b_used = ( $b->{cluster}->{bandwidth_used} * 100.0 ) / ( $b->{cluster}->{bandwidth_limit} * 1.0 );
+
+    return $a_used <=> $b_used;
+}
+
+#
+# Output DNS records for a server
+#
+sub output_server
+{
+    my $zonefile = shift;
+    my $name = shift;
+    my $cluster = shift;
+
+    foreach my $server (@{$cluster->{servers}})
+    {
+        if ($server->{status} eq "up")
+        {
+            $zonefile->print("+${name}:$server->{ipv4}:600\n");
+
+            if ($server->{ipv6})
+            {
+#                $zonefile->print("3${name}:$server->{ipv6}:600\n");
+            }
+        }
+    }
+
+    return;
+}