]> git.openstreetmap.org Git - chef.git/blobdiff - cookbooks/nominatim/recipes/default.rb
Use backport of cgi-mapserver
[chef.git] / cookbooks / nominatim / recipes / default.rb
index 30b2a20d760be5373e455e58a49569da30e32051..7eb6a1e749b5cc4f11b90237d41a269dc32f37fa 100644 (file)
 #
 
 include_recipe "accounts"
-include_recipe "munin"
-include_recipe "php::fpm"
 include_recipe "prometheus"
+include_recipe "postgresql"
+include_recipe "python"
+include_recipe "nginx"
+include_recipe "git"
+include_recipe "fail2ban"
 
 basedir = data_bag_item("accounts", "nominatim")["home"]
-email_errors = data_bag_item("accounts", "lonvia")["email"]
+project_directory = "#{basedir}/planet-project"
+bin_directory = "#{basedir}/bin"
+cfg_directory = "#{basedir}/etc"
+ui_directory = "#{basedir}/ui"
+qa_data_directory = "#{basedir}/qa-data"
 
 directory basedir do
   owner "nominatim"
@@ -32,6 +39,28 @@ directory basedir do
   recursive true
 end
 
+[basedir, bin_directory, cfg_directory, project_directory, ui_directory].each do |path|
+  directory path do
+    owner "nominatim"
+    group "nominatim"
+    mode "755"
+  end
+end
+
+if node[:nominatim][:flatnode_file]
+  directory File.dirname(node[:nominatim][:flatnode_file]) do
+    recursive true
+  end
+end
+
+directory "#{bin_directory}/maintenance" do
+  owner "nominatim"
+  group "nominatim"
+  mode "775"
+end
+
+## Log directory setup
+
 directory node[:nominatim][:logdir] do
   owner "nominatim"
   group "nominatim"
@@ -46,16 +75,7 @@ file "#{node[:nominatim][:logdir]}/query.log" do
   mode "664"
 end
 
-file "#{node[:nominatim][:logdir]}/update.log" do
-  action :create_if_missing
-  owner "nominatim"
-  group "adm"
-  mode "664"
-end
-
-## Postgresql
-
-include_recipe "postgresql"
+### Postgresql
 
 postgresql_version = node[:nominatim][:dbcluster].split("/").first
 postgis_version = node[:nominatim][:postgis]
@@ -66,325 +86,104 @@ 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
   cluster node[:nominatim][:dbcluster]
   superuser true
-  only_if { node[:nominatim][:state] != "slave" }
 end
 
 postgresql_user "www-data" do
   cluster node[:nominatim][:dbcluster]
-  only_if { node[:nominatim][:state] != "slave" }
-end
-
-postgresql_munin "nominatim" do
-  cluster node[:nominatim][:dbcluster]
-  database node[:nominatim][:dbname]
-end
-
-directory "#{basedir}/tablespaces" do
-  owner "postgres"
-  group "postgres"
-  mode "700"
-end
-
-# 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 "700"
-    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
 
-if node[:nominatim][:state] == "master"
-  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 "700"
-  end
+### Nominatim
 
-  template "/usr/local/bin/clean-db-nominatim" do
-    source "clean-db-nominatim.erb"
-    owner "root"
-    group "root"
-    mode "755"
-    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"
+python_directory = "#{basedir}/venv"
 
 package %w[
   build-essential
-  cmake
-  g++
-  libboost-dev
-  libboost-system-dev
-  libboost-filesystem-dev
-  libexpat1-dev
-  zlib1g-dev
-  libxml2-dev
-  libbz2-dev
-  libpq-dev
-  libgeos++-dev
-  libproj-dev
-  python3-pyosmium
-  pyosmium
-  python3-psycopg2
-  python3-dotenv
-  python3-psutil
-  python3-jinja2
-  python3-icu
-  python3-datrie
-  php-pgsql
-  php-intl
-  php-symfony-dotenv
+  libicu-dev
+  python3-dev
+  pkg-config
+  osm2pgsql
   ruby
   ruby-file-tail
   ruby-pg
+  ruby-webrick
 ]
 
-source_directory = "#{basedir}/nominatim"
-build_directory = "#{basedir}/bin"
-ui_directory = "#{basedir}/ui"
-qa_bin_directory = "#{basedir}/Nominatim-Data-Analyser"
-qa_data_directory = "#{basedir}/qa-data"
-
-directory build_directory do
-  owner "nominatim"
-  group "nominatim"
-  mode "755"
-  recursive true
-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
-  action node[:nominatim][:state] == "slave" ? :sync : :checkout
-  repository node[:nominatim][:repository]
-  revision node[:nominatim][:revision]
-  enable_submodules true
-  user "nominatim"
-  group "nominatim"
-  not_if { node[:nominatim][:state] != "slave" && File.exist?("#{source_directory}/README.md") }
-  notifies :run, "execute[compile_nominatim]"
+python_virtualenv python_directory do
+  interpreter "/usr/bin/python3"
 end
 
-remote_file "#{source_directory}/data/country_osm_grid.sql.gz" do
-  action :create_if_missing
-  source "https://www.nominatim.org/data/country_grid.sql.gz"
-  owner "nominatim"
-  group "nominatim"
-  mode "644"
+# These are updated during the database update.
+python_package "nominatim-db" do
+  python_virtualenv python_directory
+  extra_index_url node[:nominatim][:pip_index]
 end
 
-execute "compile_nominatim" do
-  action :nothing
-  user "nominatim"
-  cwd build_directory
-  command "cmake #{source_directory} && make"
+python_package "nominatim-api" do
+  python_virtualenv python_directory
+  extra_index_url node[:nominatim][:pip_index]
 end
 
-template "#{source_directory}/.git/hooks/post-merge" do
-  source "git-post-merge-hook.erb"
+remote_directory "#{project_directory}/static-website" do
+  source "website"
   owner "nominatim"
   group "nominatim"
   mode "755"
-  variables :srcdir => source_directory,
-            :builddir => build_directory,
-            :dbname => node[:nominatim][:dbname]
+  files_owner "nominatim"
+  files_group "nominatim"
+  files_mode "644"
+  purge false
 end
 
-template "#{build_directory}/.env" do
+template "#{project_directory}/.env" do
   source "nominatim.env.erb"
   owner "nominatim"
   group "nominatim"
   mode "664"
-  variables :base_url => node[:nominatim][:state] == "off" ? node[:fqdn] : "nominatim.openstreetmap.org",
+  variables :base_url => "nominatim.openstreetmap.org",
             :dbname => node[:nominatim][:dbname],
             :flatnode_file => node[:nominatim][:flatnode_file],
             :log_file => "#{node[:nominatim][:logdir]}/query.log",
-            :tokenizer => node[:nominatim][:config][:tokenizer]
+            :pool_size => node[:nominatim][:api_pool_size],
+            :query_timeout => node[:nominatim][:api_query_timeout],
+            :request_timeout => node[:nominatim][:api_request_timeout]
 end
 
-git ui_directory do
-  action :sync
-  repository node[:nominatim][:ui_repository]
-  revision node[:nominatim][:ui_revision]
-  user "nominatim"
+remote_file "#{project_directory}/secondary_importance.sql.gz" do
+  action :create_if_missing
+  source "https://nominatim.org/data/wikimedia-secondary-importance.sql.gz"
+  owner "nominatim"
   group "nominatim"
+  mode "644"
 end
 
-template "#{ui_directory}/dist/theme/config.theme.js" do
-  source "ui-config.js.erb"
+remote_file "#{project_directory}/wikimedia-importance.csv.gz" do
+  action :create_if_missing
+  source "https://nominatim.org/data/wikimedia-importance.csv.gz"
   owner "nominatim"
   group "nominatim"
-  mode "664"
-end
-
-if node[:nominatim][:flatnode_file]
-  directory File.dirname(node[:nominatim][:flatnode_file]) do
-    recursive true
-  end
-end
-
-template "/etc/logrotate.d/nominatim" do
-  source "logrotate.nominatim.erb"
-  owner "root"
-  group "root"
   mode "644"
 end
 
-external_data = [
-  "wikimedia-importance.sql.gz",
-  "gb_postcodes.csv.gz",
-  "us_postcodes.csv.gz"
-]
-
-external_data.each do |fname|
-  remote_file "#{build_directory}/#{fname}" do
+%w[gb_postcodes.csv.gz us_postcodes.csv.gz].each do |fname|
+  remote_file "#{project_directory}/#{fname}" do
     action :create
-    source "https://www.nominatim.org/data/#{fname}"
+    source "https://nominatim.org/data/#{fname}"
     owner "nominatim"
     group "nominatim"
     mode "644"
   end
 end
 
-if node[:nominatim][:state] == "off"
-  cron_d "nominatim-backup" do
-    action :delete
-  end
-
-  cron_d "nominatim-vacuum-db" do
-    action :delete
-  end
-
-  cron_d "nominatim-clean-db" do
-    action :delete
-  end
-
-  systemd_timer "nominatim-update-maintenance-trigger" do
-    action :delete
-  end
-else
-  cron_d "nominatim-backup" do
-    action node[:nominatim][:enable_backup] ? :create : :delete
-    minute "0"
-    hour "3"
-    day "1"
-    user "nominatim"
-    command "/usr/local/bin/backup-nominatim"
-    mailto email_errors
-  end
-
-  cron_d "nominatim-vacuum-db" do
-    minute "20"
-    hour "0"
-    user "postgres"
-    command "/usr/local/bin/vacuum-db-nominatim"
-    mailto email_errors
-  end
-
-  cron_d "nominatim-clean-db" do
-    action node[:nominatim][:state] == "master" ? :create : :delete
-    minute "5"
-    hour "*/4"
-    user "postgres"
-    command "/usr/local/bin/clean-db-nominatim"
-    mailto email_errors
-  end
-
-  systemd_service "nominatim-update-maintenance-trigger" do
-    description "Trigger maintenance tasks for Nominatim DB"
-    exec_start "touch #{basedir}/status/update_maintenance"
-    user "nominatim"
-  end
-
-  systemd_timer "nominatim-update-maintenance-trigger" do
-    action :create
-    description "Schedule maintenance tasks for Nominatim DB"
-    on_calendar "*-*-* 02:03:00 UTC"
-  end
-
-  service "nominatim-update-maintenance-trigger" do
-    action [:enable]
-  end
-end
-
-template "#{source_directory}/utils/nominatim-update" do
-  source "updater.erb"
-  user "nominatim"
-  group "nominatim"
-  mode "755"
-  variables :bindir => build_directory,
-            :srcdir => source_directory,
-            :logfile => "#{node[:nominatim][:logdir]}/update.log",
-            :branch => node[:nominatim][:revision],
-            :update_stop_file => "#{basedir}/status/updates_disabled",
-            :update_maintenance_trigger => "#{basedir}/status/update_maintenance",
-            :qabindir => qa_bin_directory,
-            :qadatadir => qa_data_directory
-end
-
-template "/etc/init.d/nominatim-update" do
-  source "updater.init.erb"
-  user "nominatim"
-  group "nominatim"
-  mode "755"
-  variables :source_directory => source_directory
-end
-
-%w[backup-nominatim vacuum-db-nominatim].each do |fname|
-  template "/usr/local/bin/#{fname}" do
-    source "#{fname}.erb"
-    owner "root"
-    group "root"
-    mode "755"
-    variables :db => node[:nominatim][:dbname]
-  end
-end
-
-## webserver frontend
-
-directory "#{basedir}/etc" do
-  owner "nominatim"
-  group "adm"
-  mode "775"
-end
+# Webserver + frontend
 
 %w[user_agent referrer email generic].each do |name|
-  file "#{basedir}/etc/nginx_blocked_#{name}.conf" do
+  file "#{cfg_directory}/nginx_blocked_#{name}.conf" do
     action :create_if_missing
     owner "nominatim"
     group "adm"
@@ -392,17 +191,26 @@ end
   end
 end
 
-node[:nominatim][:fpm_pools].each do |name, data|
-  php_fpm name do
-    port data[:port]
-    pm data[:pm]
-    pm_max_children data[:max_children]
-    pm_start_servers 20
-    pm_min_spare_servers 10
-    pm_max_spare_servers 20
-    pm_max_requests 10000
-    prometheus_port data[:prometheus_port]
-  end
+systemd_service "nominatim" do
+  description "Nominatim running as a gunicorn application"
+  user "www-data"
+  group "www-data"
+  working_directory project_directory
+  standard_output "append:#{node[:nominatim][:logdir]}/gunicorn.log"
+  standard_error "inherit"
+  exec_start "#{python_directory}/bin/gunicorn --max-requests 200000 -b unix:/run/gunicorn-nominatim.openstreetmap.org.sock -w #{node[:nominatim][:api_workers]} -k uvicorn.workers.UvicornWorker 'nominatim_api.server.falcon.server:run_wsgi()'"
+  exec_reload "/bin/kill -s HUP $MAINPID"
+  kill_mode "mixed"
+  timeout_stop_sec 5
+  private_tmp true
+  requires "nominatim.socket"
+  after "network.target"
+end
+
+systemd_socket "nominatim" do
+  description "Gunicorn socket for Nominatim"
+  listen_stream "/run/gunicorn-nominatim.openstreetmap.org.sock"
+  socket_user "www-data"
 end
 
 ssl_certificate node[:fqdn] do
@@ -417,8 +225,6 @@ ssl_certificate node[:fqdn] do
   notifies :reload, "service[nginx]"
 end
 
-include_recipe "nginx"
-
 nginx_site "default" do
   action [:delete]
 end
@@ -427,7 +233,7 @@ frontends = search(:node, "recipes:web\\:\\:frontend").sort_by(&:name)
 
 nginx_site "nominatim" do
   template "nginx.erb"
-  directory build_directory
+  directory project_directory
   variables :pools => node[:nominatim][:fpm_pools],
             :frontends => frontends,
             :confdir => "#{basedir}/etc",
@@ -441,72 +247,72 @@ template "/etc/logrotate.d/nginx" do
   mode "644"
 end
 
-munin_plugin_conf "nominatim" do
-  template "munin.erb"
-  variables :db => node[:nominatim][:dbname],
-            :querylog => "#{node[:nominatim][:logdir]}/query.log"
-end
+### Import, update and maintenance scripts
 
-munin_plugin "nominatim_importlag" do
-  target "#{source_directory}/munin/nominatim_importlag"
+%w[nominatim-update
+   nominatim-update-data
+   nominatim-update-refresh-db
+   nominatim-daily-maintenance].each do |fname|
+  template "#{bin_directory}/#{fname}" do
+    source "#{fname}.erb"
+    owner "nominatim"
+    group "nominatim"
+    mode "554"
+    variables :bindir => bin_directory,
+              :projectdir => project_directory,
+              :venvprefix => "#{python_directory}/bin/",
+              :qadatadir => qa_data_directory
+  end
 end
 
-munin_plugin "nominatim_query_speed" do
-  target "#{source_directory}/munin/nominatim_query_speed_querylog"
+systemd_service "nominatim-update" do
+  description "Update the Nominatim database"
+  exec_start "#{bin_directory}/nominatim-update"
+  restart "on-success"
+  standard_output "journal"
+  standard_error "inherit"
+  working_directory project_directory
 end
 
-munin_plugin "nominatim_requests" do
-  target "#{source_directory}/munin/nominatim_requests_querylog"
+systemd_service "nominatim-update-maintenance-trigger" do
+  description "Trigger daily maintenance tasks for Nominatim DB"
+  exec_start "ln -sf #{bin_directory}/nominatim-daily-maintenance #{bin_directory}/maintenance/"
+  user "nominatim"
 end
 
-package "ruby-webrick"
-
-prometheus_exporter "nominatim" do
-  port 8082
-  user "www-data"
-  options [
-    "--nominatim.query-log=#{node[:nominatim][:logdir]}/query.log",
-    "--nominatim.database-name=#{node[:nominatim][:dbname]}"
-  ]
+systemd_timer "nominatim-update-maintenance-trigger" do
+  action :create
+  description "Schedule daily maintenance tasks for Nominatim DB"
+  on_calendar "*-*-* 02:03:00 UTC"
 end
 
-directory "#{basedir}/status" do
-  owner "nominatim"
-  group "postgres"
-  mode "775"
+service "nominatim-update-maintenance-trigger" do
+  action :enable
 end
 
-include_recipe "fail2ban"
+## Nominatim UI
 
-frontend_addresses = frontends.collect { |f| f.ipaddresses(:role => :external) }
+git ui_directory do
+  action :sync
+  repository node[:nominatim][:ui_repository]
+  revision node[:nominatim][:ui_revision]
+  user "nominatim"
+  group "nominatim"
+end
 
-fail2ban_jail "nominatim_limit_req" do
-  filter "nginx-limit-req"
-  logpath "#{node[:nominatim][:logdir]}/nominatim.openstreetmap.org-error.log"
-  ports [80, 443]
-  maxretry 20
-  ignoreips frontend_addresses.flatten.sort
+template "#{ui_directory}/dist/theme/config.theme.js" do
+  source "ui-config.js.erb"
+  owner "nominatim"
+  group "nominatim"
+  mode "664"
 end
 
-### QA tile generation
+## Nominatim QA
 
 if node[:nominatim][:enable_qa_tiles]
-  package "python3-geojson"
-
-  git qa_bin_directory do
-    repository node[:nominatim][:qa_repository]
-    revision node[:nominatim][:qa_revision]
-    enable_submodules true
-    user "nominatim"
-    group "nominatim"
-    notifies :run, "execute[compile_qa]"
-  end
-
-  execute "compile_qa" do
-    action :nothing
-    user "nominatim"
-    cwd "#{qa_bin_directory}/clustering-vt"
-    command "make"
+  python_package "nominatim-data-analyser" do
+    python_virtualenv python_directory
+    extra_index_url node[:nominatim][:pip_index]
   end
 
   directory qa_data_directory do
@@ -516,7 +322,7 @@ if node[:nominatim][:enable_qa_tiles]
     recursive true
   end
 
-  template "#{qa_bin_directory}/analyser/config/config.yaml" do
+  template "#{project_directory}/qa-config.yaml" do
     source "qa_config.erb"
     owner "nominatim"
     group "nominatim"
@@ -531,8 +337,36 @@ if node[:nominatim][:enable_qa_tiles]
 
   nginx_site "qa-tiles.nominatim" do
     template "nginx-qa-tiles.erb"
-    directory build_directory
+    directory qa_data_directory
     variables :qa_data_directory => qa_data_directory
   end
+end
+
+## Logging and monitoring
+
+template "/etc/logrotate.d/nominatim" do
+  source "logrotate.nominatim.erb"
+  owner "root"
+  group "root"
+  mode "644"
+end
+
+prometheus_exporter "nominatim" do
+  port 8082
+  user "www-data"
+  restrict_address_families "AF_UNIX"
+  options [
+    "--nominatim.query-log=#{node[:nominatim][:logdir]}/query.log",
+    "--nominatim.database-name=#{node[:nominatim][:dbname]}"
+  ]
+end
+
+frontend_addresses = frontends.collect { |f| f.ipaddresses(:role => :external) }
 
+fail2ban_jail "nominatim_limit_req" do
+  filter "nginx-limit-req"
+  logpath "#{node[:nominatim][:logdir]}/nominatim.openstreetmap.org-error.log"
+  ports [80, 443]
+  maxretry 20
+  ignoreips frontend_addresses.flatten.sort
 end