karafka 1.1.0 → 1.2.0

data/lib/karafka/persistence/topic.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Karafka
+  module Persistence
+    # Local cache for routing topics
+    # We use it in order not to build string instances and remap incoming topic upon each
+    # message / message batches received
+    class Topic
+      # Thread.current scope under which we store topics data
+      PERSISTENCE_SCOPE = :topics
+
+      # @param group_id [String] group id for which we fetch a topic representation
+      # @param raw_topic_name [String] raw topic name (before remapping) for which we fetch a
+      #   topic representation
+      # @return [Karafka::Routing::Topic] remapped topic representation that can be used further
+      #   on when working with given parameters
+      def self.fetch(group_id, raw_topic_name)
+        Thread.current[PERSISTENCE_SCOPE] ||= Hash.new { |hash, key| hash[key] = {} }
+
+        Thread.current[PERSISTENCE_SCOPE][group_id][raw_topic_name] ||= begin
+          # We map from incoming topic name, as it might be namespaced, etc.
+          # @see topic_mapper internal docs
+          mapped_topic_name = Karafka::App.config.topic_mapper.incoming(raw_topic_name)
+          Routing::Router.find("#{group_id}_#{mapped_topic_name}")
+        end
+      end
+    end
+  end
+end

data/lib/karafka/process.rb

@@ -8,7 +8,9 @@ module Karafka
 
     # Signal types that we handle
     HANDLED_SIGNALS = %i[
-      SIGINT SIGQUIT SIGTERM
+      SIGINT
+      SIGQUIT
+      SIGTERM
     ].freeze
 
     HANDLED_SIGNALS.each do |signal|
@@ -27,8 +29,7 @@ module Karafka
 
     # Creates an instance of process and creates empty hash for callbacks
     def initialize
-      @callbacks = {}
-      HANDLED_SIGNALS.each { |signal| @callbacks[signal] = [] }
+      @callbacks = Hash.new { |hsh, key| hsh[key] = [] }
     end
 
     # Method catches all HANDLED_SIGNALS and performs appropriate callbacks (if defined)
@@ -56,7 +57,7 @@ module Karafka
     #   we have to spin up a new thread to do this
     def notice_signal(signal)
       Thread.new do
-        Karafka.monitor.notice(self.class, signal: signal)
+        Karafka.monitor.instrument('process.notice_signal', caller: self, signal: signal)
       end
     end
   end

data/lib/karafka/responders/builder.rb

@@ -3,30 +3,31 @@
 module Karafka
   # Responders namespace encapsulates all the internal responder implementation parts
   module Responders
-    # Responders builder is used to find (based on the controller class name) a responder that
-    # match the controller. This is used when user does not provide a responder inside routing
-    # but he still names responder with the same convention (and namespaces) as controller
+    # Responders builder is used to finding (based on the consumer class name) a responder
+    # that match the consumer. We use it when user does not provide a responder inside routing,
+    # but he still names responder with the same convention (and namespaces) as consumer
+    #
     # @example Matching responder exists
-    #   Karafka::Responder::Builder(NewEventsController).build #=> NewEventsResponder
+    #   Karafka::Responder::Builder(NewEventsConsumer).build #=> NewEventsResponder
     # @example Matching responder does not exist
-    #   Karafka::Responder::Builder(NewBuildsController).build #=> nil
+    #   Karafka::Responder::Builder(NewBuildsConsumer).build #=> nil
     class Builder
-      # @param controller_class [Karafka::BaseController, nil] descendant of
-      #   Karafka::BaseController
-      # @example Tries to find a responder that matches a given controller. If nothing found,
-      #   will return nil (nil is accepted, because it means that a given controller don't
+      # @param consumer_class [Karafka::BaseConsumer, nil] descendant of
+      #   Karafka::BaseConsumer
+      # @example Tries to find a responder that matches a given consumer. If nothing found,
+      #   will return nil (nil is accepted, because it means that a given consumer don't
       #   pipe stuff further on)
-      def initialize(controller_class)
-        @controller_class = controller_class
+      def initialize(consumer_class)
+        @consumer_class = consumer_class
       end
 
-      # Tries to figure out a responder based on a controller class name
+      # Tries to figure out a responder based on a consumer class name
       # @return [Class] Responder class (not an instance)
       # @return [nil] or nil if there's no matching responding class
       def build
         Helpers::ClassMatcher.new(
-          @controller_class,
-          from: 'Controller',
+          @consumer_class,
+          from: 'Consumer',
           to: 'Responder'
         ).match
       end

data/lib/karafka/routing/builder.rb

@@ -6,7 +6,7 @@ module Karafka
     # @example Build a simple (most common) route
     #   consumers do
     #     topic :new_videos do
-    #       controller NewVideosController
+    #       consumer NewVideosConsumer
     #     end
     #   end
     class Builder < Array

data/lib/karafka/routing/consumer_mapper.rb

@@ -17,7 +17,7 @@ module Karafka
     #   module MyMapper
     #     def self.call(raw_consumer_group_name)
     #       [
-    #         Karafka::App.config.client_id.to_s.underscope,
+    #         Dry::Inflector.new.underscore(Karafka::App.config.client_id.to_s),
     #         raw_consumer_group_name
     #       ].join('_').gsub('_', '.')
     #     end
@@ -26,7 +26,8 @@ module Karafka
       # @param raw_consumer_group_name [String, Symbol] string or symbolized consumer group name
       # @return [String] remapped final consumer group name
       def self.call(raw_consumer_group_name)
-        "#{Karafka::App.config.client_id.to_s.underscore}_#{raw_consumer_group_name}"
+        client_name = Dry::Inflector.new.underscore(Karafka::App.config.client_id.to_s)
+        "#{client_name}_#{raw_consumer_group_name}"
       end
     end
   end

data/lib/karafka/routing/router.rb

@@ -3,7 +3,7 @@
 module Karafka
   # Namespace for all elements related to requests routing
   module Routing
-    # Karafka framework Router for routing incoming messages to proper controllers
+    # Karafka framework Router for routing incoming messages to proper consumers
     # @note Since Kafka does not provide namespaces or modules for topics, they all have "flat"
     #  structure so all the routes are being stored in a single level array
     module Router

data/lib/karafka/routing/topic.rb

@@ -9,7 +9,7 @@ module Karafka
       extend Helpers::ConfigRetriever
 
       attr_reader :id, :consumer_group
-      attr_accessor :controller
+      attr_accessor :consumer
 
       # @param [String, Symbol] name of a topic on which we want to listen
       # @param consumer_group [Karafka::Routing::ConsumerGroup] owning consumer group of this topic
@@ -29,20 +29,14 @@ module Karafka
       # example for Sidekiq
       def build
         Karafka::AttributesMap.topic.each { |attr| send(attr) }
-        controller&.topic = self
+        consumer&.topic = self
         self
       end
 
       # @return [Class, nil] Class (not an instance) of a responder that should respond from
-      #   controller back to Kafka (usefull for piping dataflows)
+      #   consumer back to Kafka (usefull for piping dataflows)
       def responder
-        @responder ||= Karafka::Responders::Builder.new(controller).build
-      end
-
-      # @return [Class] Parser class (not instance) that we want to use to unparse Kafka messages
-      # @note If not provided - will use Json as default
-      def parser
-        @parser ||= Karafka::Parsers::Json
+        @responder ||= Karafka::Responders::Builder.new(consumer).build
       end
 
       Karafka::AttributesMap.topic.each do |attribute|
@@ -58,7 +52,7 @@ module Karafka
 
         Hash[map].merge!(
           id: id,
-          controller: controller
+          consumer: consumer
         )
       end
     end

data/lib/karafka/schemas/config.rb

@@ -13,8 +13,11 @@ module Karafka
     #   so we validate all of that once all the routes are defined and ready
     Config = Dry::Validation.Schema do
       required(:client_id).filled(:str?, format?: Karafka::Schemas::TOPIC_REGEXP)
+      required(:shutdown_timeout) { none? | (int? & gteq?(0)) }
       required(:consumer_mapper)
       required(:topic_mapper)
+      required(:params_base_class).filled
+
       optional(:backend).filled
     end
   end

data/lib/karafka/schemas/consumer_group.rb

@@ -9,6 +9,9 @@ module Karafka
       # but someetimes loads things twice
       URI_SCHEMES ||= %w[kafka kafka+ssl].freeze
 
+      # Available sasl scram mechanism of authentication (plus nil)
+      SASL_SCRAM_MECHANISMS ||= %w[sha256 sha512].freeze
+
       configure do
         config.messages_file = File.join(
           Karafka.gem_root, 'config', 'errors.yml'
@@ -21,7 +24,7 @@ module Karafka
           uri = URI.parse(uri)
           URI_SCHEMES.include?(uri.scheme) && uri.port
         rescue URI::InvalidURIError
-          return false
+          false
         end
       end
 
@@ -36,6 +39,7 @@ module Karafka
       required(:connect_timeout).filled { (int? | float?) & gt?(0) }
       required(:socket_timeout).filled { (int? | float?) & gt?(0) }
       required(:min_bytes).filled(:int?, gt?: 0)
+      required(:max_bytes).filled(:int?, gt?: 0)
       required(:max_wait_time).filled { (int? | float?) & gteq?(0) }
       required(:batch_fetching).filled(:bool?)
       required(:topics).filled { each { schema(ConsumerGroupTopic) } }
@@ -52,14 +56,22 @@ module Karafka
         ssl_ca_cert_file_path
         ssl_client_cert
         ssl_client_cert_key
+        sasl_gssapi_principal
+        sasl_gssapi_keytab
         sasl_plain_authzid
         sasl_plain_username
         sasl_plain_password
-        sasl_gssapi_principal
-        sasl_gssapi_keytab
+        sasl_scram_username
+        sasl_scram_password
       ].each do |encryption_attribute|
         optional(encryption_attribute).maybe(:str?)
       end
+
+      optional(:ssl_ca_certs_from_system).maybe(:bool?)
+
+      # It's not with other encryptions as it has some more rules
+      optional(:sasl_scram_mechanism)
+        .maybe(:str?, included_in?: Karafka::Schemas::SASL_SCRAM_MECHANISMS)
     end
   end
 end

data/lib/karafka/schemas/consumer_group_topic.rb

@@ -7,7 +7,7 @@ module Karafka
       required(:id).filled(:str?, format?: Karafka::Schemas::TOPIC_REGEXP)
       required(:name).filled(:str?, format?: Karafka::Schemas::TOPIC_REGEXP)
       required(:backend).filled(included_in?: %i[inline sidekiq])
-      required(:controller).filled
+      required(:consumer).filled
       required(:parser).filled
       required(:max_bytes_per_partition).filled(:int?, gteq?: 0)
       required(:start_from_beginning).filled(:bool?)

data/lib/karafka/server.rb

@@ -3,6 +3,13 @@
 module Karafka
   # Karafka consuming server class
   class Server
+    @consumer_threads = Concurrent::Array.new
+
+    # How long should we sleep between checks on shutting down consumers
+    SUPERVISION_SLEEP = 1
+    # What system exit code should we use when we terminated forcefully
+    FORCEFUL_EXIT_CODE = 2
+
     class << self
       # Set of consuming threads. Each consumer thread contains a single consumer
       attr_accessor :consumer_threads
@@ -12,7 +19,6 @@ module Karafka
 
       # Method which runs app
       def run
-        @consumer_threads = Concurrent::Array.new
         bind_on_sigint
         bind_on_sigquit
         bind_on_sigterm
@@ -35,17 +41,17 @@ module Karafka
 
       # What should happen when we decide to quit with sigint
       def bind_on_sigint
-        process.on_sigint { Karafka::App.stop! }
+        process.on_sigint { stop_supervised }
       end
 
       # What should happen when we decide to quit with sigquit
       def bind_on_sigquit
-        process.on_sigquit { Karafka::App.stop! }
+        process.on_sigquit { stop_supervised }
       end
 
       # What should happen when we decide to quit with sigterm
       def bind_on_sigterm
-        process.on_sigterm { Karafka::App.stop! }
+        process.on_sigterm { stop_supervised }
       end
 
       # Starts Karafka with a supervision
@@ -54,9 +60,35 @@ module Karafka
       def start_supervised
         process.supervise do
           Karafka::App.run!
-          Karafka::Fetcher.new.fetch_loop
+          Karafka::Fetcher.call
         end
       end
+
+      # Stops Karafka with a supervision (as long as there is a shutdown timeout)
+      # If consumers won't stop in a given timeframe, it will force them to exit
+      def stop_supervised
+        Karafka.monitor.instrument('server.stop', {})
+
+        Karafka::App.stop!
+        # If there is no shutdown timeout, we don't exit and wait until all the consumers
+        # had done their work
+        return unless Karafka::App.config.shutdown_timeout
+
+        # If there is a timeout, we check every 1 second (for the timeout period) if all
+        # the threads finished their work and if so, we can just return and normal
+        # shutdown process will take place
+        Karafka::App.config.shutdown_timeout.to_i.times do
+          return if consumer_threads.count(&:alive?).zero?
+          sleep SUPERVISION_SLEEP
+        end
+
+        Karafka.monitor.instrument('server.stop.error', {})
+        # We're done waiting, lets kill them!
+        consumer_threads.each(&:terminate)
+
+        # exit is not within the instrumentation as it would not trigger due to exit
+        Kernel.exit FORCEFUL_EXIT_CODE
+      end
     end
   end
 end

data/lib/karafka/setup/config.rb

@@ -13,6 +13,7 @@ module Karafka
     # @see Karafka::Setup::Configurators::Base for more details about configurators api
     class Config
       extend Dry::Configurable
+      extend Callbacks::Config
 
       # Available settings
       # option client_id [String] kafka client_id - used to provide
@@ -21,9 +22,9 @@ module Karafka
       # What backend do we want to use to process messages
       setting :backend, :inline
       # option logger [Instance] logger that we want to use
-      setting :logger, -> { ::Karafka::Logger.instance }
+      setting :logger, -> { ::Karafka::Instrumentation::Logger.instance }
       # option monitor [Instance] monitor that we will to use (defaults to Karafka::Monitor)
-      setting :monitor, -> { ::Karafka::Monitor.instance }
+      setting :monitor, -> { ::Karafka::Instrumentation::Monitor.instance }
       # Mapper used to remap consumer groups ids, so in case users migrate from other tools
       # or they need to maintain their own internal consumer group naming conventions, they
       # can easily do it, replacing the default client_id + consumer name pattern concept
@@ -34,6 +35,8 @@ module Karafka
       #   - #incoming - for remapping from the incoming message to our internal format
       #   - #outgoing - for remapping from internal topic name into outgoing message
       setting :topic_mapper, -> { Routing::TopicMapper }
+      # Default parser for parsing and unparsing incoming and outgoing data
+      setting :parser, -> { Karafka::Parsers::Json }
       # If batch_fetching is true, we will fetch kafka messages in batches instead of 1 by 1
       # @note Fetching does not equal consuming, see batch_consuming description for details
       setting :batch_fetching, true
@@ -41,12 +44,24 @@ module Karafka
       # #params_batch will contain params received from Kafka (may be more than 1) so we can
       # process them in batches
       setting :batch_consuming, false
-      # Should we operate in a single controller instance across multiple batches of messages,
-      # from the same partition or should we build a new instance for each incoming batch.
-      # Disabling that can be useful when you want to build a new controller instance for each
-      # incoming batch. It's disabled by default, not to create more objects that needed on
-      # each batch
+      # Should we operate in a single consumer instance across multiple batches of messages,
+      # from the same partition or should we build a new one for each incoming batch.
+      # Disabling that can be useful when you want to create a new consumer instance for each
+      # incoming batch. It's disabled by default, not to create more objects that needed
+      # on each batch
       setting :persistent, true
+      # option shutdown_timeout [Integer, nil] the number of seconds after which Karafka no
+      #   longer wait for the consumers to stop gracefully but instead we force
+      #   terminate everything.
+      # @note Keep in mind, that if your business logic
+      # @note If set to nil, it won't forcefully shutdown the process at all.
+      setting :shutdown_timeout, 60
+      # option params_base_class [Class] base class for params class initialization
+      #   This can be either a Hash or a HashWithIndifferentAccess depending on your
+      #   requirements. Note, that by using HashWithIndifferentAccess, you remove some of the
+      #   performance in favor of convenience. This can be useful especially if you already use
+      #   it with Rails, etc
+      setting :params_base_class, Hash
 
       # option kafka [Hash] - optional - kafka configuration options
       setting :kafka do
@@ -79,6 +94,9 @@ module Karafka
         #   returning messages from the server; if `max_wait_time` is reached, this
         #   is ignored.
         setting :min_bytes, 1
+        # option max_bytes [Integer] the maximum number of bytes to read before returning messages
+        #   from each broker.
+        setting :max_bytes, 10_485_760
         # option max_wait_time [Integer, Float] max_wait_time is the maximum number of seconds to
         #   wait before returning data from a single message fetch. By setting this high you also
         #   increase the fetching throughput - and by setting it low you set a bound on latency.
@@ -107,24 +125,33 @@ module Karafka
         setting :socket_timeout, 30
 
         # SSL authentication related settings
-        # option ca_cert [String] SSL CA certificate
+        # option ca_cert [String, nil] SSL CA certificate
         setting :ssl_ca_cert, nil
-        # option ssl_ca_cert_file_path [String] SSL CA certificate file path
+        # option ssl_ca_cert_file_path [String, nil] SSL CA certificate file path
         setting :ssl_ca_cert_file_path, nil
-        # option ssl_client_cert [String] SSL client certificate
+        # option ssl_ca_certs_from_system [Boolean] Use the CA certs from your system's default
+        #   certificate store
+        setting :ssl_ca_certs_from_system, false
+        # option ssl_client_cert [String, nil] SSL client certificate
         setting :ssl_client_cert, nil
-        # option ssl_client_cert_key [String] SSL client certificate password
+        # option ssl_client_cert_key [String, nil] SSL client certificate password
         setting :ssl_client_cert_key, nil
-        # option sasl_gssapi_principal [String] sasl principal
+        # option sasl_gssapi_principal [String, nil] sasl principal
         setting :sasl_gssapi_principal, nil
-        # option sasl_gssapi_keytab [String] sasl keytab
+        # option sasl_gssapi_keytab [String, nil] sasl keytab
         setting :sasl_gssapi_keytab, nil
         # option sasl_plain_authzid [String] The authorization identity to use
         setting :sasl_plain_authzid, ''
-        # option sasl_plain_username [String] The username used to authenticate
+        # option sasl_plain_username [String, nil] The username used to authenticate
         setting :sasl_plain_username, nil
-        # option sasl_plain_password [String] The password used to authenticate
+        # option sasl_plain_password [String, nil] The password used to authenticate
         setting :sasl_plain_password, nil
+        # option sasl_scram_username [String, nil] The username used to authenticate
+        setting :sasl_scram_username, nil
+        # option sasl_scram_password [String, nil] The password used to authenticate
+        setting :sasl_scram_password, nil
+        # option sasl_scram_mechanism [String, nil] Scram mechanism, either 'sha256' or 'sha512'
+        setting :sasl_scram_mechanism, nil
       end
 
       class << self
@@ -132,18 +159,17 @@ module Karafka
         # @yield Runs a block of code providing a config singleton instance to it
         # @yieldparam [Karafka::Setup::Config] Karafka config instance
         def setup
-          configure do |config|
-            yield(config)
-          end
+          configure { |config| yield(config) }
         end
 
         # Everything that should be initialized after the setup
         # Components are in karafka/config directory and are all loaded one by one
         # If you want to configure a next component, please add a proper file to config dir
         def setup_components
-          Configurators::Base.descendants.each do |klass|
-            klass.new(config).setup
-          end
+          [
+            Configurators::Params,
+            Configurators::WaterDrop
+          ].each { |klass| klass.setup(config) }
         end
 
         # Validate config based on ConfigurationSchema