Block more contacted sites from tiles

[chef.git] / cookbooks / tilecache / templates / default / nginx_tile.conf.erb

diff --git a/cookbooks/tilecache/templates/default/nginx_tile.conf.erb b/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
index 0d536322839046d6c546516799e322b82f5d7710..b8b0b5321f79b59fe99b91574dba64a72e940393 100644 (file)
--- a/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
+++ b/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
@@ -1,29 +1,34 @@
 # DO NOT EDIT - This file is being maintained by Chef
 
 upstream tile_cache_backend {
 # DO NOT EDIT - This file is being maintained by Chef
 
 upstream tile_cache_backend {

-  server 127.0.0.1:8080;
-  server 127.0.0.2:8080;
+  server 127.0.0.1:8080 weight=1000 max_fails=32;
+  server 127.0.0.2:8080 weight=1000 max_fails=32;

 
 

-  # Add the other caches to relieve pressure if local squid failing
+  # Add the tile_siblings caches to relieve pressure if local squid failing

   # Balancer: round-robin
   # Balancer: round-robin

+<% server_weight = 1000 -%>
+<% Array(@node[:tilecache][:tile_siblings]).each do |cache_peer| -%>

 <% @caches.each do |cache| -%>
 <% @caches.each do |cache| -%>

+<% if cache_peer == cache[:fqdn] -%>

 <% if cache[:hostname] != node[:hostname] -%>
 <% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>
 <% if cache[:hostname] != node[:hostname] -%>
 <% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>

-  server <%= address %>:80 backup; # Server <%= cache[:hostname] %>
+  server <%= address %>:80 weight=<%= server_weight %> max_fails=32 backup; # Server <%= cache[:hostname] %>
+<% server_weight -= server_weight.div(2) -%>
+<% end -%>
+<% end -%>

 <% end -%>
 <% end -%>
 <% end -%>
 
 <% end -%>
 <% end -%>
 <% end -%>
 

-  keepalive 512;
-  keepalive_requests 1024;
+  keepalive 128;

 }
 
 # Geo Map of tile caches
 geo $tile_cache {
 }
 
 # Geo Map of tile caches
 geo $tile_cache {

-  default 0;
+  default "client";

 <% @caches.each do |cache| -%>
 <% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>
 <% @caches.each do |cache| -%>
 <% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>

-  <%= address %> 1; # <%= cache[:hostname] %>
+  <%= address %> "cache"; # <%= cache[:hostname] %>

 <% end -%>
 <% end -%>
 }
 <% end -%>
 <% end -%>
 }


@@ -47,64 +52,106 @@ map $http_user_agent $approved_scraper {
 map $http_user_agent $denied_scraper {
   default                0; # Not denied
   ''                     1; # No User-Agent Set
 map $http_user_agent $denied_scraper {
   default                0; # Not denied
   ''                     1; # No User-Agent Set

-  '~^Python\-urllib\/'   1; # Library Default
-  '~^python\-requests\/' 1; # Library Default
-  '~^node\-fetch\/'      1; # Library Default
-  '~^R$'                 1; # Library Default
-  '~^Java\/'             1; # Library Default
-  '~^tiles$'             1; # Library Default
-  '~^okhttp\/'           1; # Library Default
-  '~^Microsoft-ATL-Native\/' 1; #Library Default
-  '/n software IPWorks HTTP/S Component - www.nsoftware.com' 1; #Library default
-  '~^Wget\/'             1; #Library Default
-  'C# TilesDownloader'   1; # Downloader
-  'MapDownloader'        1; # Downloader
-  '~^staticmaps'         1; # Downloader
-  'Android'              1; # Default or fake
-  'kc_android'           1; # Default or fake
-  'Mozilla/4.0'          1; # Fake
-  'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)' 1;  # Fake
-  '~^runtastic'          1; # App
-  '~^Where\ my\ children' 1; # App
-  'nossoonibusjp.android.crosswalk' 1; # App
-  'br.com.concisoti.potybus' 1; # App
+  '-'                    1;
+
+  # Library defaults
+  '~^Python\-urllib\/'   1;
+  '~^python\-requests\/' 1;
+  '~^node\-fetch\/'      1;
+  '~^R$'                 1;
+  '~^Java\/'             1;
+  '~^tiles$'             1;
+  '~^okhttp\/'           1;
+  '~^Microsoft-ATL-Native\/' 1;
+  '/n software IPWorks HTTP/S Component - www.nsoftware.com' 1;
+  '~^Wget\/'             1;
+  'java'                 1;
+
+  # Library defaults or fakes
+  'Android'              1;
+  'kc_android'           1;
+  'host'                 1;
+  '~^maptestapp'         1;
+  'Other'                1;
+  'osmdroid'             1;
+  '~^tilelive-http'      1;
+  '~^Java-http-client'   1;
+
+  # Fakes
+  'Mozilla/4.0'          1;
+  'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)' 1;
+
+  # Bulk downloaders
+  'C# TilesDownloader'   1;
+  'MapDownloader'        1;
+  '~^staticmaps'         1;
+
+  # Overusage apps
+  '~^runtastic'          1;
+  '~^Where\ my\ children' 1;
+  'nossoonibusjp.android.crosswalk' 1;
+  'br.com.concisoti.potybus' 1;

   'com.soft373.taptaxi'  1;
   'com.kradac.ktxcore'   1;
   'com.soft373.taptaxi'  1;
   'com.kradac.ktxcore'   1;

-  'ru.crowdsystems.topcontrol.knd' 1;
+  '~^ru.crowdsystems.topcontrol.knd' 1;
+  'pl.itaxi.driver'      1;
+  'net.uztaxi.driver'    1;
+  'OSMDroid/2.1 (its; rutaxi 3.28.0)' 1;
+  'com.helleniccomms.mercedes.driver' 1;
+  'ru.taximaster.www'    1;
+  'com.arobs.trackgps'   1;
+  'com.helleniccomms.asteras.driver' 1;
+
+  # Malware
+  'Agent Smith'          1;
+  # '~[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}' 1; # Fake UA

 }
 
 map $http_referer $denied_referer {
   default                          0; # Not denied
 }
 
 map $http_referer $denied_referer {
   default                          0; # Not denied

-  'http://www.openstreetmap.org/'  1; # Faked
-  'http://www.openstreetmap.org'   1; # Faked
-  'https://www.openstreetmap.org'  1; # Faked
-  'http://openstreetmap.org/'      1; # Faked
-  'http://openstreetmap.org'       1; # Faked
-  'https://openstreetmap.org'      1; # Faked
-  'http://www.osm.org/'            1; # Faked
-  'http://www.osm.org'             1; # Faked
-  'http://osm.org/'                1; # Faked
-  'http://osm.org'                 1; # Faked
-  'http://google.com'              1; # Faked
-  'http://www.google.com'          1; # Faked
-  'http://google.com/'             1; # Faked
-  'http://www.google.com/'         1; # Faked
-  'https://google.com'             1; # Faked
-  'https://www.google.com'         1; # Faked
-  'https://google.com/'            1; # Faked
-  'https://www.google.com/'        1; # Faked
-  '~^https?://pmap\.kuku\.lu/'           1; # Too much traffic
-  '~^https?://[^.]*\.pmap\.kuku\.lu/'    1; # Too much traffic
-  '~^https?://fastpokemap\.com/'         1; # Too much traffic
-  '~^https?://[^.]*\.fastpokemap\.com/'  1; # Too much traffic
-  '~^https?://pkget\.com/'               1; # Too much traffic
-  '~^https?://[^.]*\.pkget\.com/'        1; # Too much traffic
-  '~^https?://twpkinfo\.com/'            1; # Too much traffic
-  '~^https?://[^.]*\.twpkinfo\.com/'     1; # Too much traffic
-  '~^https?://9db\.jp/'                  1; # Too much traffic
-  '~^https?://[^.]*\.9db\.jp/'           1; # Too much traffic
-  '~^https?://clustrmaps\.com/'          1; # Too much traffic
-  '~^https?://[^.]*\.clustrmaps\.com/'   1; # Too much traffic
+  # Faked sites
+  'http://www.openstreetmap.org/'  1;
+  'http://www.openstreetmap.org'   1;
+  'https://www.openstreetmap.org'  1;
+  'http://openstreetmap.org/'      1;
+  'http://openstreetmap.org'       1;
+  'https://openstreetmap.org'      1;
+  'http://www.osm.org/'            1;
+  'http://www.osm.org'             1;
+  'http://osm.org/'                1;
+  'http://osm.org'                 1;
+  'http://google.com'              1;
+  'http://www.google.com'          1;
+  'http://google.com/'             1;
+  'http://www.google.com/'         1;
+  'https://google.com'             1;
+  'https://www.google.com'         1;
+  'https://google.com/'            1;
+  'https://www.google.com/'        1;
+  'http://www.microsoft.com/'      1;
+
+  # Overusing websites
+  '~^https?://pmap\.kuku\.lu/'           1;
+  '~^https?://[^.]*\.pmap\.kuku\.lu/'    1;
+  '~^https?://fastpokemap\.com/'         1;
+  '~^https?://[^.]*\.fastpokemap\.com/'  1;
+  '~^https?://pkget\.com/'               1;
+  '~^https?://[^.]*\.pkget\.com/'        1;
+  '~^https?://twpkinfo\.com/'            1;
+  '~^https?://[^.]*\.twpkinfo\.com/'     1;
+  '~^https?://9db\.jp/'                  1;
+  '~^https?://[^.]*\.9db\.jp/'           1;
+  '~^https?://clustrmaps\.com/'          1;
+  '~^https?://[^.]*\.clustrmaps\.com/'   1;
+  '~^https?://geoportal360\.pl/'         1;
+  '~^https?://skelbiu\.lt/'              1;
+  '~^https?://[^.]*\.skelbiu\.lt/'       1;
+  '~^https?://wialon\.[^.]*/'            1; # wialon has many domains, so block them all
+  '~^https?://[^.]*\.wialon\.[^.]*/'     1;
+  '~^https?://gps-trace\.com/'           1;
+  '~^https?://[^.]*\.gps-trace\.com/'    1;
+  '~^https?://cellmapper\.net/'          1;
+  '~^https?://[^.]*\.cellmapper\.net/'   1;

 }
 
 map $http_referer $osm_referer {
 }
 
 map $http_referer $osm_referer {


@@ -113,17 +160,34 @@ map $http_referer $osm_referer {
 }
 
 # Limit Cache-Control header to only approved User-Agents
 }
 
 # Limit Cache-Control header to only approved User-Agents

-map $osm_referer$http_user_agent $limit_http_cache_control {
-  default '';                                # Unset Header
-  '~^osmMozilla\/5\.0\ QGIS\/' '';            # Unset Header
-  '~^osmMozilla\/5\.0\ ' $http_cache_control; # Pass Header
+map $tile_cache$osm_referer$http_user_agent $limit_http_cache_control {
+  default '';                             # Unset Header
+  '~^clientosmMozilla\/5\.0\ \(X11'       $http_cache_control; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Windows'   $http_cache_control; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(iPhone'    $http_cache_control; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Macintosh' $http_cache_control; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Linux'     $http_cache_control; # Pass Header

 }
 
 # Limit Pragma header to only approved User-Agents
 }
 
 # Limit Pragma header to only approved User-Agents

-map $osm_referer$http_user_agent $limit_http_pragma {
-  default '';                          # Unset Header
-  '~^osmMozilla\/5\.0\ QGIS\/' '';     # Unset Header
-  '~^osmMozilla\/5\.0\ ' $http_pragma; # Pass Header
+map $tile_cache$osm_referer$http_user_agent $limit_http_pragma {
+  default '';                             # Unset Header
+  '~^clientosmMozilla\/5\.0\ \(X11'       $http_pragma; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Windows'   $http_pragma; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(iPhone'    $http_pragma; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Macintosh' $http_pragma; # Pass Header
+  '~^clientosmMozilla\/5\.0\ \(Linux'     $http_pragma; # Pass Header
+}
+
+# Find Browser User-Agents which are not sending a referer.
+# Browsers with no referer could be due to Browser extension or website Referrer-Policy
+map $tile_cache$http_referer$scheme$http_user_agent $deny_missing_referer {
+  default                                   0; # Not denied
+  '~^clienthttpsMozilla\/5\.0\ \(X11'       1;
+  '~^clienthttpsMozilla\/5\.0\ \(Windows'   1;
+  '~^clienthttpsMozilla\/5\.0\ \(iPhone'    1;
+  '~^clienthttpsMozilla\/5\.0\ \(Macintosh' 1;
+  '~^clienthttpsMozilla\/5\.0\ \(Linux'     1;

 }
 
 server {
 }
 
 server {


@@ -148,66 +212,51 @@ server {
     # Immediately 404 layers we do not support
 <% for i in 20..99 do %>
     location /<%= i %>/ {
     # Immediately 404 layers we do not support
 <% for i in 20..99 do %>
     location /<%= i %>/ {

-      set $limit_rate 512;

       return 404;
     }
 <% end %>
 
     # Immediately 404 silly tile requests
     location = /0/0/-1.png {
       return 404;
     }
 <% end %>
 
     # Immediately 404 silly tile requests
     location = /0/0/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/0/-1.png {
       return 404;
     }
     location = /1/0/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/-1/0.png {
       return 404;
     }
     location = /1/-1/0.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/-1/1.png {
       return 404;
     }
     location = /1/-1/1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/-1/-1.png {
       return 404;
     }
     location = /1/-1/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/-1/2.png {
       return 404;
     }
     location = /1/-1/2.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/1/-1.png {
       return 404;
     }
     location = /1/1/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /1/2/-1.png {
       return 404;
     }
     location = /1/2/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/0/-1.png {
       return 404;
     }
     location = /2/0/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/-1/0.png {
       return 404;
     }
     location = /2/-1/0.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/-1/1.png {
       return 404;
     }
     location = /2/-1/1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/1/-1.png {
       return 404;
     }
     location = /2/1/-1.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/-1/2.png {
       return 404;
     }
     location = /2/-1/2.png {

-      set $limit_rate 512;

       return 404;
     }
     location = /2/-1/3.png {
       return 404;
     }
     location = /2/-1/3.png {

-      set $limit_rate 512;

       return 404;
     }
 
       return 404;
     }
 


@@ -231,6 +280,8 @@ server {
 
       proxy_connect_timeout 20s;
 
 
       proxy_connect_timeout 20s;
 

+      proxy_next_upstream_tries 2;
+

       # Replace host header.
       proxy_set_header Host 'tile.openstreetmap.org';
       # Do not pass cookies to backends.
       # Replace host header.
       proxy_set_header Host 'tile.openstreetmap.org';
       # Do not pass cookies to backends.


@@ -273,7 +324,14 @@ server {
       proxy_cache_revalidate on;
       proxy_cache_min_uses 4;
 
       proxy_cache_revalidate on;
       proxy_cache_min_uses 4;
 

-      add_header x-cache-status $upstream_cache_status;
+      add_header x-cache-status "$upstream_cache_status - <%= node[:hostname] %>";
+<% else %>
+      # Severely rate limit Browser UAs which are not sending a referer.
+      # With no referer we do not know who is using tiles
+      if ($deny_missing_referer) {
+        set $limit_rate 1024;
+        add_header x-cache-ratelimit "missing-referer";
+      }

 <% end -%>
 
       # Set a QoS cookie if none presented (uses nginx Map)
 <% end -%>
 
       # Set a QoS cookie if none presented (uses nginx Map)