X-Git-Url: https://git.openstreetmap.org./chef.git/blobdiff_plain/36690d4cd2b53ecb17cc2302a3552b6c6d05c4cb..71782b10c75d33947d536a9c8adccb039601155d:/cookbooks/prometheus/recipes/server.rb

diff --git a/cookbooks/prometheus/recipes/server.rb b/cookbooks/prometheus/recipes/server.rb
index 11ff21dab..4b9058c8e 100644
--- a/cookbooks/prometheus/recipes/server.rb
+++ b/cookbooks/prometheus/recipes/server.rb
@@ -18,129 +18,127 @@
 #
 
 include_recipe "apache"
-include_recipe "apt"
+include_recipe "apt::grafana"
+include_recipe "awscli"
 include_recipe "networking"
-include_recipe "timescaledb"
 
 passwords = data_bag_item("prometheus", "passwords")
 tokens = data_bag_item("prometheus", "tokens")
+admins = data_bag_item("apache", "admins")
 
 prometheus_exporter "fastly" do
   port 8080
-  listen_switch "endpoint"
-  listen_type "url"
+  listen_switch "listen"
   environment "FASTLY_API_TOKEN" => tokens["fastly"]
 end
 
-package %w[
-  prometheus
-  prometheus-alertmanager
-]
-
-promscale_version = "0.6.1"
+prometheus_exporter "fastly_healthcheck" do
+  port 9696
+  scrape_interval "1m"
+  environment "FASTLY_API_TOKEN" => tokens["fastly"]
+end
 
-database_cluster = "#{node[:timescaledb][:database_version]}/main"
+prometheus_exporter "statuscake" do
+  port 9595
+  scrape_interval "5m"
+  scrape_timeout "2m"
+  environment "STATUSCAKE_APIKEY" => tokens["statuscake"]
+end
 
-postgresql_user "prometheus" do
-  cluster database_cluster
-  superuser true
+template "/etc/prometheus/cloudwatch.yml" do
+  source "cloudwatch.yml.erb"
+  owner "root"
+  group "root"
+  mode "644"
 end
 
-postgresql_database "promscale" do
-  cluster database_cluster
-  owner "prometheus"
+prometheus_exporter "cloudwatch" do
+  address "127.0.0.1"
+  port 5000
+  listen_switch "listen-address"
+  options %w[
+    --config.file=/etc/prometheus/cloudwatch.yml
+    --enable-feature=aws-sdk-v2
+    --enable-feature=always-return-info-metrics
+  ]
+  environment "AWS_ACCESS_KEY_ID" => "AKIASQUXHPE7JHG37EA6",
+              "AWS_SECRET_ACCESS_KEY" => tokens["cloudwatch"]
+  subscribes :restart, "template[/etc/prometheus/cloudwatch.yml]"
 end
 
-directory "/opt/promscale" do
+cache_dir = Chef::Config[:file_cache_path]
+
+prometheus_version = "2.45.0"
+alertmanager_version = "0.25.0"
+karma_version = "0.114"
+
+directory "/opt/prometheus-server" do
   owner "root"
   group "root"
   mode "755"
 end
 
-package %w[
-  make
-  gcc
-  clang-9
-  llvm-9
-  cargo
-]
+prometheus_arch = if arm?
+                    "arm64"
+                  else
+                    "amd64"
+                  end
 
-git "/opt/promscale/extension" do
-  action :sync
-  repository "https://github.com/timescale/promscale_extension.git"
-  revision "0.2.0"
-  user "root"
+remote_file "#{cache_dir}/prometheus.linux.tar.gz" do
+  source "https://github.com/prometheus/prometheus/releases/download/v#{prometheus_version}/prometheus-#{prometheus_version}.linux-#{prometheus_arch}.tar.gz"
+  owner "root"
   group "root"
+  mode "644"
+  backup false
 end
 
-execute "/opt/promscale/extension/Makefile" do
+archive_file "#{cache_dir}/prometheus.linux.tar.gz" do
   action :nothing
-  command "make install"
-  cwd "/opt/promscale/extension"
-  user "root"
+  destination "/opt/prometheus-server/prometheus"
+  overwrite true
+  strip_components 1
+  owner "root"
   group "root"
-  subscribes :run, "git[/opt/promscale/extension]", :immediately
-  notifies :restart, "service[postgresql]", :immediately
+  subscribes :extract, "remote_file[#{cache_dir}/prometheus.linux.tar.gz]", :immediately
 end
 
-directory "/opt/promscale/bin" do
+remote_file "#{cache_dir}/alertmanager.linux.tar.gz" do
+  source "https://github.com/prometheus/alertmanager/releases/download/v#{alertmanager_version}/alertmanager-#{alertmanager_version}.linux-#{prometheus_arch}.tar.gz"
   owner "root"
   group "root"
-  mode "755"
+  mode "644"
+  backup false
 end
 
-remote_file "/opt/promscale/bin/promscale" do
-  action :create
-  source "https://github.com/timescale/promscale/releases/download/#{promscale_version}/promscale_#{promscale_version}_Linux_x86_64"
+archive_file "#{cache_dir}/alertmanager.linux.tar.gz" do
+  action :nothing
+  destination "/opt/prometheus-server/alertmanager"
+  overwrite true
+  strip_components 1
   owner "root"
   group "root"
-  mode "755"
-end
-
-systemd_service "promscale" do
-  description "Promscale Connector"
-  type "simple"
-  user "prometheus"
-  exec_start "/opt/promscale/bin/promscale --db-uri postgresql:///promscale?host=/run/postgresql&port=5432 --db-connections-max 400"
-  # exec_start lazy { "/opt/promscale/bin/promscale --db-host /run/postgresql --db-port #{node[:postgresql][:clusters][database_cluster][:port]} --db-user prometheus --db-name promscale --db-max-connections 400" }
-  limit_nofile 16384
-  private_tmp true
-  protect_system "strict"
-  protect_home true
-  no_new_privileges true
+  subscribes :extract, "remote_file[#{cache_dir}/alertmanager.linux.tar.gz]", :immediately
 end
 
-service "promscale" do
-  action [:enable, :start]
-  subscribes :restart, "remote_file[/opt/promscale/bin/promscale]"
-  subscribes :restart, "systemd_service[promscale]"
-end
-
-systemd_service "promscale-maintenance" do
-  description "Promscale Maintenance"
-  type "simple"
-  user "prometheus"
-  exec_start "/usr/bin/psql --command='CALL prom_api.execute_maintenance()' promscale"
-  private_tmp true
-  protect_system "strict"
-  protect_home true
-  no_new_privileges true
-end
-
-systemd_timer "promscale-maintenance" do
-  description "Promscale Maintenace"
-  on_active_sec 1800
-  on_unit_inactive_sec 1800
+remote_file "#{cache_dir}/karma-linux.tar.gz" do
+  source "https://github.com/prymitive/karma/releases/download/v#{karma_version}/karma-linux-#{prometheus_arch}.tar.gz"
+  owner "root"
+  group "root"
+  mode "644"
+  backup false
 end
 
-service "promscale-maintenance.timer" do
-  action [:enable, :start]
+archive_file "#{cache_dir}/karma-linux.tar.gz" do
+  action :nothing
+  destination "/opt/prometheus-server/karma"
+  overwrite true
+  owner "root"
+  group "root"
+  subscribes :extract, "remote_file[#{cache_dir}/karma-linux.tar.gz]", :immediately
 end
 
 search(:node, "roles:gateway") do |gateway|
-  allowed_ips = gateway.interfaces(:role => :internal).map do |interface|
-    "#{interface[:network]}/#{interface[:prefix]}"
-  end
+  allowed_ips = gateway.ipaddresses(:role => :internal).map(&:subnet)
 
   node.default[:networking][:wireguard][:peers] << {
     :public_key => gateway[:networking][:wireguard][:public_key],
@@ -150,6 +148,7 @@ search(:node, "roles:gateway") do |gateway|
 end
 
 jobs = {}
+junos_targets = []
 snmp_targets = []
 
 search(:node, "recipes:prometheus\\:\\:default").sort_by(&:name).each do |client|
@@ -165,26 +164,47 @@ search(:node, "recipes:prometheus\\:\\:default").sort_by(&:name).each do |client
     if exporter.is_a?(Hash)
       name = exporter[:name]
       address = exporter[:address]
+      sni = exporter[:sni]
+      labels = Array(exporter[:labels])
+      scrape_interval = exporter[:scrape_interval]
+      scrape_timeout = exporter[:scrape_timeout]
       metric_relabel = exporter[:metric_relabel] || []
     else
       name = key
       address = exporter
+      sni = nil
+      labels = []
+      scrape_interval = nil
+      scrape_timeout = nil
       metric_relabel = []
     end
 
     jobs[name] ||= []
     jobs[name] << {
       :address => address,
+      :sni => sni,
       :instance => client.name.split(".").first,
+      :labels => labels,
+      :scrape_interval => scrape_interval,
+      :scrape_timeout => scrape_timeout,
       :metric_relabel => metric_relabel
     }
   end
 
+  Hash(client[:prometheus][:junos]).each do |instance, details|
+    junos_targets << {
+      :instance => instance,
+      :target => details[:address],
+      :address => client[:prometheus][:addresses]["junos"],
+      :labels => Array(details[:labels])
+    }
+  end
+
   Hash(client[:prometheus][:snmp]).each do |instance, details|
     snmp_targets << {
       :instance => instance,
       :target => details[:address],
-      :module => details[:module],
+      :modules => details[:modules],
       :address => client[:prometheus][:addresses]["snmp"],
       :labels => Array(details[:labels])
     }
@@ -217,11 +237,14 @@ prometheus_exporter "ssl" do
   register_target false
 end
 
-template "/etc/default/prometheus" do
-  source "default.prometheus.erb"
-  owner "root"
-  group "root"
-  mode "644"
+package "prometheus"
+
+systemd_service "prometheus-executable" do
+  service "prometheus"
+  dropin "executable"
+  exec_start "/opt/prometheus-server/prometheus/prometheus --config.file=/etc/prometheus/prometheus.yml --web.enable-admin-api --web.external-url=https://prometheus.openstreetmap.org/prometheus --storage.tsdb.path=/var/lib/prometheus/metrics2 --storage.tsdb.retention.time=540d"
+  timeout_stop_sec 300
+  notifies :restart, "service[prometheus]"
 end
 
 template "/etc/prometheus/prometheus.yml" do
@@ -229,7 +252,7 @@ template "/etc/prometheus/prometheus.yml" do
   owner "root"
   group "root"
   mode "644"
-  variables :jobs => jobs, :snmp_targets => snmp_targets, :certificates => certificates
+  variables :jobs => jobs, :junos_targets => junos_targets, :snmp_targets => snmp_targets, :certificates => certificates
 end
 
 template "/etc/prometheus/alert_rules.yml" do
@@ -241,16 +264,24 @@ end
 
 service "prometheus" do
   action [:enable, :start]
-  subscribes :restart, "template[/etc/default/prometheus]"
   subscribes :reload, "template[/etc/prometheus/prometheus.yml]"
   subscribes :reload, "template[/etc/prometheus/alert_rules.yml]"
+  subscribes :restart, "archive_file[#{cache_dir}/prometheus.linux.tar.gz]"
 end
 
-template "/etc/default/prometheus-alertmanager" do
-  source "default.alertmanager.erb"
-  owner "root"
-  group "root"
-  mode "644"
+systemd_service "prometheus-alertmanager" do
+  description "Prometheus alert manager"
+  type "simple"
+  user "prometheus"
+  exec_start "/opt/prometheus-server/alertmanager/alertmanager --config.file=/etc/prometheus/alertmanager.yml --storage.path=/var/lib/prometheus/alertmanager --web.external-url=https://prometheus.openstreetmap.org/alertmanager"
+  exec_reload "/bin/kill -HUP $MAINPID"
+  timeout_stop_sec 20
+  restart "on-failure"
+  notifies :restart, "service[prometheus-alertmanager]"
+end
+
+link "/usr/local/bin/promtool" do
+  to "/opt/prometheus-server/prometheus/promtool"
 end
 
 template "/etc/prometheus/alertmanager.yml" do
@@ -260,19 +291,58 @@ template "/etc/prometheus/alertmanager.yml" do
   mode "644"
 end
 
+directory "/var/lib/prometheus/alertmanager" do
+  owner "prometheus"
+  group "prometheus"
+  mode "755"
+end
+
 service "prometheus-alertmanager" do
   action [:enable, :start]
-  subscribes :restart, "template[/etc/default/prometheus-alertmanager]"
   subscribes :reload, "template[/etc/prometheus/alertmanager.yml]"
+  subscribes :restart, "systemd_service[prometheus-alertmanager]"
+  subscribes :restart, "archive_file[#{cache_dir}/alertmanager.linux.tar.gz]"
+end
+
+directory "/etc/amtool" do
+  owner "root"
+  group "root"
+  mode "755"
 end
 
-template "/etc/prometheus/amtool.yml" do
+template "/etc/amtool/config.yml" do
   source "amtool.yml.erb"
   owner "root"
   group "root"
   mode "644"
 end
 
+link "/usr/local/bin/amtool" do
+  to "/opt/prometheus-server/alertmanager/amtool"
+end
+
+template "/etc/prometheus/karma.yml" do
+  source "karma.yml.erb"
+  owner "root"
+  group "root"
+  mode "644"
+end
+
+systemd_service "prometheus-karma" do
+  description "Alert dashboard for Prometheus Alertmanager"
+  user "prometheus"
+  exec_start "/opt/prometheus-server/karma/karma-linux-#{prometheus_arch} --config.file=/etc/prometheus/karma.yml"
+  sandbox :enable_network => true
+  restart "on-failure"
+end
+
+service "prometheus-karma" do
+  action [:enable, :start]
+  subscribes :restart, "template[/etc/prometheus/karma.yml]"
+  subscribes :restart, "archive_file[#{cache_dir}/karma-linux.tar.gz]"
+  subscribes :restart, "systemd_service[prometheus-karma]"
+end
+
 package "grafana-enterprise"
 
 template "/etc/grafana/grafana.ini" do
@@ -290,14 +360,16 @@ end
 
 apache_module "alias"
 apache_module "proxy_http"
+apache_module "proxy_wstunnel"
 
 ssl_certificate "prometheus.openstreetmap.org" do
-  domains ["prometheus.openstreetmap.org", "prometheus.osm.org"]
+  domains ["prometheus.openstreetmap.org", "prometheus.osm.org", "munin.openstreetmap.org", "munin.osm.org"]
   notifies :reload, "service[apache2]"
 end
 
 apache_site "prometheus.openstreetmap.org" do
   template "apache.erb"
+  variables :admin_hosts => admins["hosts"]
 end
 
 template "/etc/cron.daily/prometheus-backup" do
@@ -306,3 +378,49 @@ template "/etc/cron.daily/prometheus-backup" do
   group "root"
   mode "750"
 end
+
+package %w[
+  curl
+  jq
+]
+
+directory "/var/lib/prometheus/.aws" do
+  user "prometheus"
+  group "prometheus"
+  mode "755"
+end
+
+template "/var/lib/prometheus/.aws/credentials" do
+  source "aws-credentials.erb"
+  user "prometheus"
+  group "prometheus"
+  mode "600"
+  variables :passwords => passwords
+end
+
+template "/usr/local/bin/prometheus-backup-data" do
+  source "backup-data.erb"
+  owner "root"
+  group "root"
+  mode "755"
+end
+
+systemd_service "prometheus-backup-data" do
+  description "Backup prometheus data to S3"
+  user "prometheus"
+  exec_start "/usr/local/bin/prometheus-backup-data"
+  read_write_paths %w[
+    /var/lib/prometheus/.aws
+    /var/lib/prometheus/metrics2/snapshots
+  ]
+  sandbox :enable_network => true
+end
+
+systemd_timer "prometheus-backup-data" do
+  description "Backup prometheus data to S3"
+  on_calendar "03:11"
+end
+
+service "prometheus-backup-data.timer" do
+  action [:enable, :start]
+end