nominatim: add read timeout to apache

[chef.git] / cookbooks / nominatim / recipes / default.rb

diff --git a/cookbooks/nominatim/recipes/default.rb b/cookbooks/nominatim/recipes/default.rb
index 44806430032eec74ad8ce650d3565ea7780ac092..a8009759ff3dba77926ab82870173bf54b2d8b5e 100644 (file)
--- a/cookbooks/nominatim/recipes/default.rb
+++ b/cookbooks/nominatim/recipes/default.rb
@@ -1,8 +1,8 @@
 #
 # Cookbook Name:: nominatim
 #
 # Cookbook Name:: nominatim

-# Recipe:: default
+# Recipe:: base

 #
 #

-# Copyright 2012, OpenStreetMap Foundation
+# Copyright 2015, OpenStreetMap Foundation

 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.


@@ -17,223 +17,362 @@
 # limitations under the License.
 #
 
 # limitations under the License.
 #
 

-include_recipe "apache::ssl"
-include_recipe "postgresql"
-include_recipe "git"
-
-package "php5"
-package "php5-cli"
-package "php5-pgsql"
-package "php5-fpm"
-package "php-pear"
-package "php-apc"
-
-apache_module "rewrite"
-apache_module "fastcgi-handler"
-
-home_directory = data_bag_item("accounts", "nominatim")["home"]
-source_directory = "#{home_directory}/nominatim"
+basedir = data_bag_item("accounts", "nominatim")["home"]

 email_errors = data_bag_item("accounts", "lonvia")["email"]
 
 email_errors = data_bag_item("accounts", "lonvia")["email"]
 

-service "php5-fpm" do
-  action [ :enable, :start ]
-  supports :status => true, :restart => true, :reload => true
+directory basedir do
+  owner "nominatim"
+  group "nominatim"
+  mode 0o755
+  recursive true

 end
 
 end
 

-apache_site "nominatim.openstreetmap.org" do
-  template "apache.erb"
-  directory source_directory
-  variables :pools => node[:nominatim][:fpm_pools]
+directory node[:nominatim][:logdir] do
+  owner "nominatim"
+  group "nominatim"
+  mode 0o755
+  recursive true

 end
 
 end
 

-node[:nominatim][:fpm_pools].each do |name,data|
-
-  template "/etc/php5/fpm/pool.d/#{name}.conf" do
-    source "fpm.conf.erb"
-    owner "root"
-    group "root"
-    mode 0644
-    variables data.merge(:name => name)
-    notifies :reload, "service[php5-fpm]"
-  end
+file "#{node[:nominatim][:logdir]}/query.log" do
+  action :create_if_missing
+  owner "www-data"
+  group "adm"
+  mode 0o664

 end
 
 end
 

-postgresql_user "tomh" do
-  cluster "9.1/main"
-  superuser true
+file "#{node[:nominatim][:logdir]}/update.log" do
+  action :create_if_missing
+  owner "nominatim"
+  group "adm"
+  mode 0o664

 end
 
 end
 

-postgresql_user "lonvia" do
-  cluster "9.1/main"
-  superuser true
-end
+## Postgresql

 
 

-postgresql_user "twain" do
-  cluster "9.1/main"
-  superuser true
+include_recipe "postgresql"
+
+package "postgis"
+
+node[:nominatim][:dbadmins].each do |user|
+  postgresql_user user do
+    cluster node[:nominatim][:dbcluster]
+    superuser true
+    only_if { node[:nominatim][:state] != "slave" }
+  end

 end
 
 postgresql_user "nominatim" do
 end
 
 postgresql_user "nominatim" do

-  cluster "9.1/main"
+  cluster node[:nominatim][:dbcluster]

   superuser true
   superuser true

+  only_if { node[:nominatim][:state] != "slave" }

 end
 
 postgresql_user "www-data" do
 end
 
 postgresql_user "www-data" do

-  cluster "9.1/main"
+  cluster node[:nominatim][:dbcluster]
+  only_if { node[:nominatim][:state] != "slave" }

 end
 
 postgresql_munin "nominatim" do
 end
 
 postgresql_munin "nominatim" do

-  cluster "9.1/main"
-  database "nominatim"
+  cluster node[:nominatim][:dbcluster]
+  database node[:nominatim][:dbname]

 end
 
 end
 

-directory "/var/log/nominatim" do
-  owner "nominatim"
-  group "nominatim"
-  mode 0755
+directory "#{basedir}/tablespaces" do
+  owner "postgres"
+  group "postgres"
+  mode 0o700

 end
 
 end
 

-template "/etc/logrotate.d/nominatim" do
-  source "logrotate.nominatim.erb"
-  owner "root"
-  group "root"
-  mode 0644
+# Note: tablespaces must be exactly in the same location on each
+#       Nominatim instance when replication is in use. Therefore
+#       use symlinks to canonical directory locations.
+node[:nominatim][:tablespaces].each do |name, location|
+  directory location do
+    owner "postgres"
+    group "postgres"
+    mode 0o700
+    recursive true
+  end
+
+  link "#{basedir}/tablespaces/#{name}" do
+    to location
+  end
+
+  postgresql_tablespace name do
+    cluster node[:nominatim][:dbcluster]
+    location "#{basedir}/tablespaces/#{name}"
+  end

 end
 
 end
 

+if node[:nominatim][:state] == "master" # ~FC023
+  postgresql_user "replication" do
+    cluster node[:nominatim][:dbcluster]
+    password data_bag_item("nominatim", "passwords")["replication"]
+    replication true
+  end
+
+  directory node[:rsyncd][:modules][:archive][:path] do
+    owner "postgres"
+    group "postgres"
+    mode 0o700
+  end
+
+  template "/usr/local/bin/clean-db-nominatim" do
+    source "clean-db-nominatim.erb"
+    owner "root"
+    group "root"
+    mode 0o755
+    variables :archive_dir => node[:rsyncd][:modules][:archive][:path],
+              :update_stop_file => "#{basedir}/status/updates_disabled",
+              :streaming_clients => search(:node, "nominatim_state:slave").map { |slave| slave[:fqdn] }.join(" ")
+  end
+end
+
+## Nominatim backend
+
+include_recipe "git"

 
 

-package "osmosis"
-package "gcc"
-package "proj-bin"
-package "libgeos-c1"
-package "postgresql-9.1-postgis"
-package "postgresql-server-dev-9.1"

 package "build-essential"
 package "build-essential"

+package "cmake"
+package "g++"
+package "libboost-dev"
+package "libboost-system-dev"
+package "libboost-filesystem-dev"
+package "libexpat1-dev"
+package "zlib1g-dev"

 package "libxml2-dev"
 package "libxml2-dev"

-package "libgeos-dev"
-package "libgeos++-dev"
-package "libpq-dev"

 package "libbz2-dev"
 package "libbz2-dev"

-package "libtool"
-package "automake"
+package "libpq-dev"
+package "libgeos++-dev"

 package "libproj-dev"
 package "libproj-dev"

-package "libprotobuf-c0-dev"
-package "protobuf-c-compiler"
-package "python-psycopg2"
+package "osmosis"

 
 

-execute "php-pear-db" do
-  command "pear install DB"
-  not_if { File.exists?("/usr/share/php/DB") }
-end
+source_directory = "#{basedir}/nominatim"
+build_directory = "#{basedir}/bin"

 
 

-execute "compile_nominatim" do
-  action :nothing
-  command "cd #{source_directory} && ./autogen.sh && ./configure && make"
-  user "nominatim"
+directory build_directory do
+  owner "nominatim"
+  group "nominatim"
+  mode 0o755
+  recursive true

 end
 
 end
 

+# Normally syncing via chef is a bad idea because syncing might involve
+# an update of database functions which should not be done while an update
+# is ongoing. Therefore we sync in between update cycles. There is an
+# exception for slaves: they get DB function updates from the master, so
+# only the source code needs to be updated, which chef may do.

 git source_directory do
 git source_directory do

-  action :checkout
+  action node[:nominatim][:state] == "slave" ? :sync : :checkout

   repository node[:nominatim][:repository]
   repository node[:nominatim][:repository]

+  revision node[:nominatim][:revision]

   enable_submodules true
   user "nominatim"
   group "nominatim"
   enable_submodules true
   user "nominatim"
   group "nominatim"

-  notifies :run, "execute[compile_nominatim]"
+  notifies :run, "execute[compile_nominatim]", :immediately
+end
+
+execute "compile_nominatim" do
+  action :nothing
+  user "nominatim"
+  cwd build_directory
+  command "cmake #{source_directory} && make"
+end
+
+template "#{source_directory}/.git/hooks/post-merge" do
+  source "git-post-merge-hook.erb"
+  owner "nominatim"
+  group "nominatim"
+  mode 0o755
+  variables :srcdir => source_directory,
+            :builddir => build_directory,
+            :dbname => node[:nominatim][:dbname]

 end
 
 end
 

-directory "#{source_directory}/log" do
+template "#{build_directory}/settings/local.php" do
+  source "settings.erb"

   owner "nominatim"
   group "nominatim"
   owner "nominatim"
   group "nominatim"

-  mode 0755
+  mode 0o664
+  variables :base_url => node[:nominatim][:state] == "off" ? node[:fqdn] : "nominatim.openstreetmap.org",
+            :dbname => node[:nominatim][:dbname],
+            :flatnode_file => node[:nominatim][:flatnode_file],
+            :log_file => "#{node[:nominatim][:logdir]}/query.log"

 end
 
 end
 

+if node[:nominatim][:flatnode_file] # ~FC023
+  directory File.dirname(node[:nominatim][:flatnode_file]) do
+    recursive true
+  end
+end

 
 

-template "#{source_directory}/.git/hooks/post-merge" do
-  source "update_source.erb"
-  owner  "nominatim"
-  group  "nominatim"
-  mode   0755
-  variables :source_directory => source_directory
+template "/etc/logrotate.d/nominatim" do
+  source "logrotate.nominatim.erb"
+  owner "root"
+  group "root"
+  mode 0o644
+end
+
+external_data = [
+  "wikipedia_article.sql.bin",
+  "wikipedia_redirect.sql.bin",
+  "gb_postcode_data.sql.gz"
+]
+
+external_data.each do |fname|
+  remote_file "#{source_directory}/data/#{fname}" do
+    action :create_if_missing
+    source "http://www.nominatim.org/data/#{fname}"
+    owner "nominatim"
+    group "nominatim"
+    mode 0o644
+  end

 end
 
 end
 

-template "#{source_directory}/settings/local.php" do
-  source "nominatim.erb"
-  owner "nominatim"
+template "/etc/cron.d/nominatim" do
+  action node[:nominatim][:state] == :off ? :delete : :create
+  source "nominatim.cron.erb"
+  owner "root"
+  group "root"
+  mode "0644"
+  variables :bin_directory => "#{source_directory}/utils", :mailto => email_errors
+end
+
+template "#{source_directory}/utils/nominatim-update" do
+  source "updater.erb"
+  user "nominatim"
+  group "nominatim"
+  mode 0o755
+  variables :bindir => build_directory,
+            :srcdir => source_directory,
+            :logfile => "#{node[:nominatim][:logdir]}/update.log",
+            :branch => node[:nominatim][:revision],
+            :update_stop_file => "#{basedir}/status/updates_disabled"
+end
+
+template "/etc/init.d/nominatim-update" do
+  source "updater.init.erb"
+  user "nominatim"

   group "nominatim"
   group "nominatim"

-  mode 0664
+  mode 0o755
+  variables :source_directory => source_directory

 end
 
 end
 

-template "#{source_directory}/settings/ip_blocks.conf" do
+%w(backup-nominatim vacuum-db-nominatim).each do |fname|
+  template "/usr/local/bin/#{fname}" do
+    source "#{fname}.erb"
+    owner "root"
+    group "root"
+    mode 0o755
+    variables :db => node[:nominatim][:dbname]
+  end
+end
+
+## webserver frontend
+
+template "#{build_directory}/settings/ip_blocks.conf" do

   action :create_if_missing
   source "ipblocks.erb"
   owner "nominatim"
   group "nominatim"
   action :create_if_missing
   source "ipblocks.erb"
   owner "nominatim"
   group "nominatim"

-  mode 0664
+  mode 0o664

 end
 
 end
 

-file "#{source_directory}/settings/apache_blocks.conf" do
+file "#{build_directory}/settings/apache_blocks.conf" do

   action :create_if_missing
   owner "nominatim"
   group "nominatim"
   action :create_if_missing
   owner "nominatim"
   group "nominatim"

-  mode 0664
+  mode 0o664

 end
 
 end
 

-file "#{source_directory}/settings/ip_blocks.map" do
+file "#{build_directory}/settings/ip_blocks.map" do

   action :create_if_missing
   owner "nominatim"
   group "nominatim"
   action :create_if_missing
   owner "nominatim"
   group "nominatim"

-  mode 0664
+  mode 0o664

 end
 
 end
 

-cron "nominatim_logrotate" do
-  hour "5"
-  minute "30"
-  weekday "0"
-  command "#{source_directory}/utils/cron_logrotate.sh"
-  user "nominatim"
-  mailto email_errors
+include_recipe "apache"
+
+package "php"
+package "php-fpm"
+package "php-pgsql"
+package "php-pear"
+package "php-db"
+
+apache_module "rewrite"
+apache_module "proxy"
+apache_module "proxy_fcgi"
+apache_module "proxy_http"
+apache_module "headers"
+apache_module "reqtimeout"
+
+service "php7.0-fpm" do
+  action [:enable, :start]
+  supports :status => true, :restart => true, :reload => true

 end
 
 end
 

-cron "nominatim_banip" do
-  command "#{source_directory}/utils/cron_banip.py"
-  user "nominatim"
-  mailto email_errors
+node[:nominatim][:fpm_pools].each do |name, data|
+  template "/etc/php/7.0/fpm/pool.d/#{name}.conf" do
+    source "fpm.conf.erb"
+    owner "root"
+    group "root"
+    mode 0o644
+    variables data.merge(:name => name)
+    notifies :reload, "service[php7.0-fpm]"
+  end

 end
 
 end
 

-cron "nominatim_vacuum" do
-  hour "2"
-  minute "00"
-  command "#{source_directory}/utils/cron_vacuum.sh"
-  user "nominatim"
-  mailto email_errors
+ssl_certificate "nominatim.openstreetmap.org" do
+  domains ["nominatim.openstreetmap.org",
+           "nominatim.osm.org",
+           "nominatim.openstreetmap.com",
+           "nominatim.openstreetmap.net",
+           "nominatim.openstreetmaps.org",
+           "nominatim.openmaps.org"]
+  notifies :reload, "service[apache2]"

 end
 
 end
 

-['search', 'reverse'].each do |filename|
-  ['phpj', 'phpx'].each do |ext|
-    link "#{source_directory}/website/#{filename}.#{ext}" do
-      to "#{source_directory}/website/#{filename}.php"
-      user "nominatim"
-      group "nominatim"
-    end
-  end
+apache_site "nominatim.openstreetmap.org" do
+  template "apache.erb"
+  directory build_directory
+  variables :pools => node[:nominatim][:fpm_pools]

 end
 
 end
 

-template "#{source_directory}/utils/nominatim-update" do
-  source "updater.erb"
-  user   "nominatim"
-  group  "nominatim"
-  mode   0755
+apache_site "default" do
+  action [:disable]

 end
 
 end
 

-template "/etc/init.d/nominatim-update" do
-  source "updater.init.erb"
-  user   "nominatim"
-  group  "nominatim"
-  mode   0755
-  variables :source_directory => source_directory
+template "/etc/logrotate.d/apache2" do
+  source "logrotate.apache.erb"
+  owner "root"
+  group "root"
+  mode 0o644
+end
+
+include_recipe "fail2ban"
+
+web_servers = search(:node, "recipes:web\\:\\:frontend").collect do |n| # ~FC010
+  n.ipaddresses(:role => :external)
+end.flatten
+
+fail2ban_filter "nominatim" do
+  failregex '^<HOST> - - \[\] "[^"]+" (408|429) '
+end
+
+fail2ban_jail "nominatim" do
+  filter "nominatim"
+  logpath "/var/log/apache2/nominatim.openstreetmap.org-access.log"
+  ports [80, 443]
+  maxretry 100
+  ignoreips web_servers

 end
 
 munin_plugin_conf "nominatim" do
   template "munin.erb"
 end
 
 munin_plugin_conf "nominatim" do
   template "munin.erb"

+  variables :db => node[:nominatim][:dbname],
+            :querylog => "#{node[:nominatim][:logdir]}/query.log"

 end
 
 munin_plugin "nominatim_importlag" do
 end
 
 munin_plugin "nominatim_importlag" do


@@ -241,29 +380,19 @@ munin_plugin "nominatim_importlag" do
 end
 
 munin_plugin "nominatim_query_speed" do
 end
 
 munin_plugin "nominatim_query_speed" do

-  target "#{source_directory}/munin/nominatim_query_speed"
+  target "#{source_directory}/munin/nominatim_query_speed_querylog"

 end
 
 munin_plugin "nominatim_requests" do
 end
 
 munin_plugin "nominatim_requests" do

-  target "#{source_directory}/munin/nominatim_requests"
+  target "#{source_directory}/munin/nominatim_requests_querylog"

 end
 
 munin_plugin "nominatim_throttled_ips" do
   target "#{source_directory}/munin/nominatim_throttled_ips"
 end
 
 end
 
 munin_plugin "nominatim_throttled_ips" do
   target "#{source_directory}/munin/nominatim_throttled_ips"
 end
 

-template "/usr/local/bin/backup-nominatim" do
-  source "backup-nominatim.erb"
-  owner "root"
-  group "root"
-  mode 0755
-end
-
-cron "nominatim_backup" do
-  hour "3"
-  minute "00"
-  day "1"
-  command "/usr/local/bin/backup-nominatim"
-  user "nominatim"
-  mailto email_errors
+directory "#{basedir}/status" do
+  owner "nominatim"
+  group "postgres"
+  mode 0o775

 end
 end