services/api/config/arvados_config.rb

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 #
   6 # Load Arvados configuration from /etc/arvados/config.yml, using defaults
   7 # from config.default.yml
   8 #
   9 # Existing application.yml is migrated into the new config structure.
  10 # Keys in the legacy application.yml take precedence.
  11 #
  12 # Use "bundle exec config:dump" to get the complete active configuration
  13 #
  14 # Use "bundle exec config:migrate" to migrate application.yml and
  15 # database.yml to config.yml.  After adding the output of
  16 # config:migrate to /etc/arvados/config.yml, you will be able to
  17 # delete application.yml and database.yml.
  18
  19 require "cgi"
  20 require 'config_loader'
  21 require 'open3'
  22
  23 begin
  24   # If secret_token.rb exists here, we need to load it first.
  25   require_relative 'secret_token.rb'
  26 rescue LoadError
  27   # Normally secret_token.rb is missing and the secret token is
  28   # configured by application.yml (i.e., here!) instead.
  29 end
  30
  31 # Load the defaults, used by config:migrate and fallback loading
  32 # legacy application.yml
  33 defaultYAML, stderr, status = Open3.capture3("arvados-server", "config-dump", "-config=-", "-skip-legacy", stdin_data: "Clusters: {xxxxx: {}}")
  34 if !status.success?
  35   puts stderr
  36   raise "error loading config: #{status}"
  37 end
  38 confs = YAML.load(defaultYAML, deserialize_symbols: false)
  39 clusterID, clusterConfig = confs["Clusters"].first
  40 $arvados_config_defaults = clusterConfig
  41 $arvados_config_defaults["ClusterID"] = clusterID
  42
  43 if ENV["ARVADOS_CONFIG"] == "none"
  44   # Don't load config. This magic value is set by packaging scripts so
  45   # they can run "rake assets:precompile" without a real config.
  46   $arvados_config_global = $arvados_config_defaults.deep_dup
  47 else
  48   # Load the global config file
  49   Open3.popen2("arvados-server", "config-dump", "-skip-legacy") do |stdin, stdout, status_thread|
  50     confs = YAML.load(stdout, deserialize_symbols: false)
  51     if confs && !confs.empty?
  52       # config-dump merges defaults with user configuration, so every
  53       # key should be set.
  54       clusterID, clusterConfig = confs["Clusters"].first
  55       $arvados_config_global = clusterConfig
  56       $arvados_config_global["ClusterID"] = clusterID
  57     else
  58       # config-dump failed, assume we will be loading from legacy
  59       # application.yml, initialize with defaults.
  60       $arvados_config_global = $arvados_config_defaults.deep_dup
  61     end
  62   end
  63 end
  64
  65 # Now make a copy
  66 $arvados_config = $arvados_config_global.deep_dup
  67
  68 def arrayToHash cfg, k, v
  69   val = {}
  70   v.each do |entry|
  71     val[entry.to_s] = {}
  72   end
  73   ConfigLoader.set_cfg cfg, k, val
  74 end
  75
  76 # Declare all our configuration items.
  77 arvcfg = ConfigLoader.new
  78 arvcfg.declare_config "ClusterID", NonemptyString, :uuid_prefix
  79 arvcfg.declare_config "ManagementToken", String, :ManagementToken
  80 arvcfg.declare_config "SystemRootToken", String
  81 arvcfg.declare_config "Git.Repositories", String, :git_repositories_dir
  82 arvcfg.declare_config "API.DisabledAPIs", Hash, :disable_api_methods, ->(cfg, k, v) { arrayToHash cfg, "API.DisabledAPIs", v }
  83 arvcfg.declare_config "API.MaxRequestSize", Integer, :max_request_size
  84 arvcfg.declare_config "API.MaxIndexDatabaseRead", Integer, :max_index_database_read
  85 arvcfg.declare_config "API.MaxItemsPerResponse", Integer, :max_items_per_response
  86 arvcfg.declare_config "API.MaxTokenLifetime", ActiveSupport::Duration
  87 arvcfg.declare_config "API.AsyncPermissionsUpdateInterval", ActiveSupport::Duration, :async_permissions_update_interval
  88 arvcfg.declare_config "Users.AutoSetupNewUsers", Boolean, :auto_setup_new_users
  89 arvcfg.declare_config "Users.AutoSetupNewUsersWithVmUUID", String, :auto_setup_new_users_with_vm_uuid
  90 arvcfg.declare_config "Users.AutoSetupNewUsersWithRepository", Boolean, :auto_setup_new_users_with_repository
  91 arvcfg.declare_config "Users.AutoSetupUsernameBlacklist", Hash, :auto_setup_name_blacklist, ->(cfg, k, v) { arrayToHash cfg, "Users.AutoSetupUsernameBlacklist", v }
  92 arvcfg.declare_config "Users.NewUsersAreActive", Boolean, :new_users_are_active
  93 arvcfg.declare_config "Users.AutoAdminUserWithEmail", String, :auto_admin_user
  94 arvcfg.declare_config "Users.AutoAdminFirstUser", Boolean, :auto_admin_first_user
  95 arvcfg.declare_config "Users.UserProfileNotificationAddress", String, :user_profile_notification_address
  96 arvcfg.declare_config "Users.AdminNotifierEmailFrom", String, :admin_notifier_email_from
  97 arvcfg.declare_config "Users.EmailSubjectPrefix", String, :email_subject_prefix
  98 arvcfg.declare_config "Users.UserNotifierEmailFrom", String, :user_notifier_email_from
  99 arvcfg.declare_config "Users.NewUserNotificationRecipients", Hash, :new_user_notification_recipients, ->(cfg, k, v) { arrayToHash cfg, "Users.NewUserNotificationRecipients", v }
 100 arvcfg.declare_config "Users.NewInactiveUserNotificationRecipients", Hash, :new_inactive_user_notification_recipients, method(:arrayToHash)
 101 arvcfg.declare_config "Login.LoginCluster", String
 102 arvcfg.declare_config "Login.TrustedClients", Hash
 103 arvcfg.declare_config "Login.RemoteTokenRefresh", ActiveSupport::Duration
 104 arvcfg.declare_config "Login.TokenLifetime", ActiveSupport::Duration
 105 arvcfg.declare_config "TLS.Insecure", Boolean, :sso_insecure
 106 arvcfg.declare_config "AuditLogs.MaxAge", ActiveSupport::Duration, :max_audit_log_age
 107 arvcfg.declare_config "AuditLogs.MaxDeleteBatch", Integer, :max_audit_log_delete_batch
 108 arvcfg.declare_config "AuditLogs.UnloggedAttributes", Hash, :unlogged_attributes, ->(cfg, k, v) { arrayToHash cfg, "AuditLogs.UnloggedAttributes", v }
 109 arvcfg.declare_config "SystemLogs.MaxRequestLogParamsSize", Integer, :max_request_log_params_size
 110 arvcfg.declare_config "Collections.DefaultReplication", Integer, :default_collection_replication
 111 arvcfg.declare_config "Collections.DefaultTrashLifetime", ActiveSupport::Duration, :default_trash_lifetime
 112 arvcfg.declare_config "Collections.CollectionVersioning", Boolean, :collection_versioning
 113 arvcfg.declare_config "Collections.PreserveVersionIfIdle", ActiveSupport::Duration, :preserve_version_if_idle
 114 arvcfg.declare_config "Collections.TrashSweepInterval", ActiveSupport::Duration, :trash_sweep_interval
 115 arvcfg.declare_config "Collections.BlobSigningKey", String, :blob_signing_key
 116 arvcfg.declare_config "Collections.BlobSigningTTL", ActiveSupport::Duration, :blob_signature_ttl
 117 arvcfg.declare_config "Collections.BlobSigning", Boolean, :permit_create_collection_with_unsigned_manifest, ->(cfg, k, v) { ConfigLoader.set_cfg cfg, "Collections.BlobSigning", !v }
 118 arvcfg.declare_config "Collections.ForwardSlashNameSubstitution", String
 119 arvcfg.declare_config "Containers.SupportedDockerImageFormats", Hash, :docker_image_formats, ->(cfg, k, v) { arrayToHash cfg, "Containers.SupportedDockerImageFormats", v }
 120 arvcfg.declare_config "Containers.LogReuseDecisions", Boolean, :log_reuse_decisions
 121 arvcfg.declare_config "Containers.DefaultKeepCacheRAM", Integer, :container_default_keep_cache_ram
 122 arvcfg.declare_config "Containers.MaxDispatchAttempts", Integer, :max_container_dispatch_attempts
 123 arvcfg.declare_config "Containers.MaxRetryAttempts", Integer, :container_count_max
 124 arvcfg.declare_config "Containers.UsePreemptibleInstances", Boolean, :preemptible_instances
 125 arvcfg.declare_config "Containers.MaxComputeVMs", Integer, :max_compute_nodes
 126 arvcfg.declare_config "Containers.Logging.LogBytesPerEvent", Integer, :crunch_log_bytes_per_event
 127 arvcfg.declare_config "Containers.Logging.LogSecondsBetweenEvents", ActiveSupport::Duration, :crunch_log_seconds_between_events
 128 arvcfg.declare_config "Containers.Logging.LogThrottlePeriod", ActiveSupport::Duration, :crunch_log_throttle_period
 129 arvcfg.declare_config "Containers.Logging.LogThrottleBytes", Integer, :crunch_log_throttle_bytes
 130 arvcfg.declare_config "Containers.Logging.LogThrottleLines", Integer, :crunch_log_throttle_lines
 131 arvcfg.declare_config "Containers.Logging.LimitLogBytesPerJob", Integer, :crunch_limit_log_bytes_per_job
 132 arvcfg.declare_config "Containers.Logging.LogPartialLineThrottlePeriod", ActiveSupport::Duration, :crunch_log_partial_line_throttle_period
 133 arvcfg.declare_config "Containers.Logging.LogUpdatePeriod", ActiveSupport::Duration, :crunch_log_update_period
 134 arvcfg.declare_config "Containers.Logging.LogUpdateSize", Integer, :crunch_log_update_size
 135 arvcfg.declare_config "Containers.Logging.MaxAge", ActiveSupport::Duration, :clean_container_log_rows_after
 136 arvcfg.declare_config "Containers.SLURM.Managed.DNSServerConfDir", Pathname, :dns_server_conf_dir
 137 arvcfg.declare_config "Containers.SLURM.Managed.DNSServerConfTemplate", Pathname, :dns_server_conf_template
 138 arvcfg.declare_config "Containers.SLURM.Managed.DNSServerReloadCommand", String, :dns_server_reload_command
 139 arvcfg.declare_config "Containers.SLURM.Managed.DNSServerUpdateCommand", String, :dns_server_update_command
 140 arvcfg.declare_config "Containers.SLURM.Managed.ComputeNodeDomain", String, :compute_node_domain
 141 arvcfg.declare_config "Containers.SLURM.Managed.ComputeNodeNameservers", Hash, :compute_node_nameservers, ->(cfg, k, v) { arrayToHash cfg, "Containers.SLURM.Managed.ComputeNodeNameservers", v }
 142 arvcfg.declare_config "Containers.SLURM.Managed.AssignNodeHostname", String, :assign_node_hostname
 143 arvcfg.declare_config "Containers.JobsAPI.Enable", String, :enable_legacy_jobs_api, ->(cfg, k, v) { ConfigLoader.set_cfg cfg, "Containers.JobsAPI.Enable", v.to_s }
 144 arvcfg.declare_config "Containers.JobsAPI.GitInternalDir", String, :git_internal_dir
 145 arvcfg.declare_config "Mail.MailchimpAPIKey", String, :mailchimp_api_key
 146 arvcfg.declare_config "Mail.MailchimpListID", String, :mailchimp_list_id
 147 arvcfg.declare_config "Services.Controller.ExternalURL", URI
 148 arvcfg.declare_config "Services.Workbench1.ExternalURL", URI, :workbench_address
 149 arvcfg.declare_config "Services.Websocket.ExternalURL", URI, :websocket_address
 150 arvcfg.declare_config "Services.WebDAV.ExternalURL", URI, :keep_web_service_url
 151 arvcfg.declare_config "Services.GitHTTP.ExternalURL", URI, :git_repo_https_base
 152 arvcfg.declare_config "Services.GitSSH.ExternalURL", URI, :git_repo_ssh_base, ->(cfg, k, v) { ConfigLoader.set_cfg cfg, "Services.GitSSH.ExternalURL", "ssh://#{v}" }
 153 arvcfg.declare_config "RemoteClusters", Hash, :remote_hosts, ->(cfg, k, v) {
 154   h = if cfg["RemoteClusters"] then
 155         cfg["RemoteClusters"].deep_dup
 156       else
 157         {}
 158       end
 159   v.each do |clusterid, host|
 160     if h[clusterid].nil?
 161       h[clusterid] = {
 162         "Host" => host,
 163         "Proxy" => true,
 164         "Scheme" => "https",
 165         "Insecure" => false,
 166         "ActivateUsers" => false
 167       }
 168     end
 169   end
 170   ConfigLoader.set_cfg cfg, "RemoteClusters", h
 171 }
 172 arvcfg.declare_config "RemoteClusters.*.Proxy", Boolean, :remote_hosts_via_dns
 173
 174 dbcfg = ConfigLoader.new
 175
 176 dbcfg.declare_config "PostgreSQL.ConnectionPool", Integer, :pool
 177 dbcfg.declare_config "PostgreSQL.Connection.host", String, :host
 178 dbcfg.declare_config "PostgreSQL.Connection.port", String, :port
 179 dbcfg.declare_config "PostgreSQL.Connection.user", String, :username
 180 dbcfg.declare_config "PostgreSQL.Connection.password", String, :password
 181 dbcfg.declare_config "PostgreSQL.Connection.dbname", String, :database
 182 dbcfg.declare_config "PostgreSQL.Connection.template", String, :template
 183 dbcfg.declare_config "PostgreSQL.Connection.encoding", String, :encoding
 184 dbcfg.declare_config "PostgreSQL.Connection.collation", String, :collation
 185
 186 application_config = {}
 187 %w(application.default application).each do |cfgfile|
 188   path = "#{::Rails.root.to_s}/config/#{cfgfile}.yml"
 189   confs = ConfigLoader.load(path, erb: true)
 190   # Ignore empty YAML file:
 191   next if confs == false
 192   application_config.deep_merge!(confs['common'] || {})
 193   application_config.deep_merge!(confs[::Rails.env.to_s] || {})
 194 end
 195
 196 db_config = {}
 197 path = "#{::Rails.root.to_s}/config/database.yml"
 198 if !ENV['ARVADOS_CONFIG_NOLEGACY'] && File.exist?(path)
 199   db_config = ConfigLoader.load(path, erb: true)
 200 end
 201
 202 $remaining_config = arvcfg.migrate_config(application_config, $arvados_config)
 203 dbcfg.migrate_config(db_config[::Rails.env.to_s] || {}, $arvados_config)
 204
 205 if application_config[:auto_activate_users_from]
 206   application_config[:auto_activate_users_from].each do |cluster|
 207     if $arvados_config.RemoteClusters[cluster]
 208       $arvados_config.RemoteClusters[cluster]["ActivateUsers"] = true
 209     end
 210   end
 211 end
 212
 213 if application_config[:host] || application_config[:port] || application_config[:scheme]
 214   if !application_config[:host] || application_config[:host].empty?
 215     raise "Must set 'host' when setting 'port' or 'scheme'"
 216   end
 217   $arvados_config.Services["Controller"]["ExternalURL"] = URI((application_config[:scheme] || "https")+"://"+application_config[:host]+
 218                                                               (if application_config[:port] then ":#{application_config[:port]}" else "" end))
 219 end
 220
 221 # Checks for wrongly typed configuration items, coerces properties
 222 # into correct types (such as Duration), and optionally raise error
 223 # for essential configuration that can't be empty.
 224 arvcfg.coercion_and_check $arvados_config_defaults, check_nonempty: false
 225 arvcfg.coercion_and_check $arvados_config_global, check_nonempty: false
 226 arvcfg.coercion_and_check $arvados_config, check_nonempty: true
 227 dbcfg.coercion_and_check $arvados_config, check_nonempty: true
 228
 229 # * $arvados_config_defaults is the defaults
 230 # * $arvados_config_global is $arvados_config_defaults merged with the contents of /etc/arvados/config.yml
 231 # These are used by the rake config: tasks
 232 #
 233 # * $arvados_config is $arvados_config_global merged with the migrated contents of application.yml
 234 # This is what actually gets copied into the Rails configuration object.
 235
 236 if $arvados_config["Collections"]["DefaultTrashLifetime"] < 86400.seconds then
 237   raise "default_trash_lifetime is %d, must be at least 86400" % Rails.configuration.Collections.DefaultTrashLifetime
 238 end
 239
 240 #
 241 # Special case for test database where there's no database.yml,
 242 # because the Arvados config.yml doesn't have a concept of multiple
 243 # rails environments.
 244 #
 245 if ::Rails.env.to_s == "test" && db_config["test"].nil?
 246   $arvados_config["PostgreSQL"]["Connection"]["dbname"] = "arvados_test"
 247 end
 248 if ::Rails.env.to_s == "test"
 249   # Use template0 when creating a new database. Avoids
 250   # character-encoding/collation problems.
 251   $arvados_config["PostgreSQL"]["Connection"]["template"] = "template0"
 252   # Some test cases depend on en_US.UTF-8 collation.
 253   $arvados_config["PostgreSQL"]["Connection"]["collation"] = "en_US.UTF-8"
 254 end
 255
 256 if ENV["ARVADOS_CONFIG"] == "none"
 257   # We need the postgresql connection URI to be valid, even if we
 258   # don't use it.
 259   $arvados_config["PostgreSQL"]["Connection"]["host"] = "localhost"
 260   $arvados_config["PostgreSQL"]["Connection"]["user"] = "x"
 261   $arvados_config["PostgreSQL"]["Connection"]["password"] = "x"
 262   $arvados_config["PostgreSQL"]["Connection"]["dbname"] = "x"
 263 end
 264
 265 if $arvados_config["PostgreSQL"]["Connection"]["password"].empty?
 266   raise "Database password is empty, PostgreSQL section is: #{$arvados_config["PostgreSQL"]}"
 267 end
 268
 269 dbhost = $arvados_config["PostgreSQL"]["Connection"]["host"]
 270 if $arvados_config["PostgreSQL"]["Connection"]["port"] != 0
 271   dbhost += ":#{$arvados_config["PostgreSQL"]["Connection"]["port"]}"
 272 end
 273
 274 #
 275 # If DATABASE_URL is set, then ActiveRecord won't error out if database.yml doesn't exist.
 276 #
 277 # For config migration, we've previously populated the PostgreSQL
 278 # section of the config from database.yml
 279 #
 280 database_url = "postgresql://#{CGI.escape $arvados_config["PostgreSQL"]["Connection"]["user"]}:"+
 281                       "#{CGI.escape $arvados_config["PostgreSQL"]["Connection"]["password"]}@"+
 282                       "#{dbhost}/#{CGI.escape $arvados_config["PostgreSQL"]["Connection"]["dbname"]}?"+
 283                       "template=#{$arvados_config["PostgreSQL"]["Connection"]["template"]}&"+
 284                       "encoding=#{$arvados_config["PostgreSQL"]["Connection"]["client_encoding"]}&"+
 285                       "collation=#{$arvados_config["PostgreSQL"]["Connection"]["collation"]}&"+
 286                       "pool=#{$arvados_config["PostgreSQL"]["ConnectionPool"]}"
 287
 288 ENV["DATABASE_URL"] = database_url
 289
 290 Server::Application.configure do
 291   # Copy into the Rails config object.  This also turns Hash into
 292   # OrderedOptions so that application code can use
 293   # Rails.configuration.API.Blah instead of
 294   # Rails.configuration.API["Blah"]
 295   ConfigLoader.copy_into_config $arvados_config, config
 296   ConfigLoader.copy_into_config $remaining_config, config
 297
 298   # We don't rely on cookies for authentication, so instead of
 299   # requiring a signing key in config, we assign a new random one at
 300   # startup.
 301   secrets.secret_key_base = rand(1<<255).to_s(36)
 302 end