rocketjob 5.4.0.beta1 → 6.0.0.rc2

data/lib/rocket_job/lookup_collection.rb

@@ -0,0 +1,69 @@
+module RocketJob
+  class LookupCollection < Mongo::Collection
+    # Rapidly upload individual records in batches.
+    #
+    # Operates directly on a Mongo Collection to avoid the overhead of creating Mongoid objects
+    # for each and every row.
+    #
+    # Example:
+    #   lookup_collection(:my_lookup).upload do |io|
+    #     io << {id: 123, data: "first record"}
+    #     io << {id: 124, data: "second record"}
+    #   end
+    #
+    #   input_category(:my_lookup).find(id: 123).first
+    def upload(batch_size: 10_000, &block)
+      BatchUploader.upload(batch_size: batch_size, &block)
+    end
+
+    # Looks up the value at the specified id.
+    # Returns [nil] if no record was found with the supplied id.
+    def lookup(id)
+      find(id: id).first
+    end
+
+    # Internal class for uploading records in batches
+    class BatchUploader
+      attr_reader :record_count
+
+      def self.upload(collection, **args)
+        writer = new(collection, **args)
+        yield(writer)
+        writer.record_count
+      ensure
+        writer&.close
+      end
+
+      def initialize(collection, batch_size:)
+        @batch_size   = batch_size
+        @record_count = 0
+        @batch_count  = 0
+        @documents    = []
+        @collection   = collection
+      end
+
+      def <<(record)
+        raise(ArgumentError, "Record must be a Hash") unless record.is_a?(Hash)
+
+        unless record.key?(:id) || record.key?("id") || record.key?("_id")
+          raise(ArgumentError, "Record must include an :id key")
+        end
+
+        @documents << record
+        @record_count += 1
+        @batch_count  += 1
+        if @batch_count >= @batch_size
+          @collection.insert_many(@documents)
+          @documents.clear
+          @batch_count = 0
+        end
+
+        self
+      end
+
+      def close
+        @collection.insert_many(@documents) unless @documents.empty?
+      end
+    end
+  end
+end

data/lib/rocket_job/plugins/job/model.rb

@@ -37,12 +37,10 @@ module RocketJob
           # arrives, then the current job will complete the current slices and process
           # the new higher priority job
           field :priority, type: Integer, default: 50, class_attribute: true, user_editable: true, copy_on_restart: true
+          validates_inclusion_of :priority, in: 1..100
 
           # When the job completes destroy it from both the database and the UI
-          field :destroy_on_complete, type: Boolean, default: true, class_attribute: true, copy_on_restart: true
-
-          # Whether to store the results from this job
-          field :collect_output, type: Boolean, default: false, class_attribute: true
+          field :destroy_on_complete, type: Mongoid::Boolean, default: true, class_attribute: true, copy_on_restart: true
 
           # Run this job no earlier than this time
           field :run_at, type: Time, user_editable: true
@@ -54,14 +52,15 @@ module RocketJob
           # Can be used to reduce log noise, especially during high volume calls
           # For debugging a single job can be logged at a low level such as :trace
           #   Levels supported: :trace, :debug, :info, :warn, :error, :fatal
-          field :log_level, type: Symbol, class_attribute: true, user_editable: true, copy_on_restart: true
+          field :log_level, type: Mongoid::StringifiedSymbol, class_attribute: true, user_editable: true, copy_on_restart: true
+          validates_inclusion_of :log_level, in: SemanticLogger::LEVELS + [nil]
 
           #
           # Read-only attributes
           #
 
           # Current state, as set by the state machine. Do not modify this value directly.
-          field :state, type: Symbol, default: :queued
+          field :state, type: Mongoid::StringifiedSymbol, default: :queued
 
           # When the job was created
           field :created_at, type: Time, default: -> { Time.now }
@@ -89,17 +88,12 @@ module RocketJob
           # Store the last exception for this job
           embeds_one :exception, class_name: "RocketJob::JobException"
 
-          # Store the Hash result from this job if collect_output is true,
-          # and the job returned actually returned a Hash, otherwise nil
-          # Not applicable to SlicedJob jobs, since its output is stored in a
-          # separate collection
-          field :result, type: Hash
-
+          # Used when workers fetch jobs to work on.
           index({state: 1, priority: 1, _id: 1}, background: true)
+          # Used by Mission Control to display completed jobs sorted by completion.
+          index({completed_at: 1}, background: true)
 
           validates_presence_of :state, :failure_count, :created_at
-          validates :priority, inclusion: 1..100
-          validates :log_level, inclusion: SemanticLogger::LEVELS + [nil]
         end
 
         module ClassMethods
@@ -155,14 +149,8 @@ module RocketJob
 
           # Scope for queued jobs that can run now
           # I.e. Queued jobs excluding scheduled jobs
-          if Mongoid::VERSION.to_f >= 7.1
-            def queued_now
-              queued.and(RocketJob::Job.where(run_at: nil).or(:run_at.lte => Time.now))
-            end
-          else
-            def queued_now
-              queued.or({run_at: nil}, :run_at.lte => Time.now)
-            end
+          def queued_now
+            queued.and(RocketJob::Job.where(run_at: nil).or(:run_at.lte => Time.now))
           end
 
           # Defines all the fields that are accessible on the Document
@@ -183,43 +171,30 @@ module RocketJob
           #
           # @return [ Field ] The generated field
           def field(name, options)
-            if options.delete(:user_editable) == true
-              self.user_editable_fields += [name.to_sym] unless user_editable_fields.include?(name.to_sym)
+            if (options.delete(:user_editable) == true) && !user_editable_fields.include?(name.to_sym)
+              self.user_editable_fields += [name.to_sym]
             end
+
             if options.delete(:class_attribute) == true
               class_attribute(name, instance_accessor: false)
               public_send("#{name}=", options[:default]) if options.key?(:default)
               options[:default] = -> { self.class.public_send(name) }
             end
-            if options.delete(:copy_on_restart) == true
-              self.rocket_job_restart_attributes += [name.to_sym] unless rocket_job_restart_attributes.include?(name.to_sym)
+
+            if (options.delete(:copy_on_restart) == true) && !rocket_job_restart_attributes.include?(name.to_sym)
+              self.rocket_job_restart_attributes += [name.to_sym]
             end
-            super(name, options)
-          end
 
-          # DEPRECATED
-          def rocket_job
-            warn "Replace calls to .rocket_job with calls to set class instance variables. For example: self.priority = 50"
-            yield(self)
+            super(name, options)
           end
 
-          # DEPRECATED
-          def public_rocket_job_properties(*args)
-            warn "Replace calls to .public_rocket_job_properties by adding `user_editable: true` option to the field declaration in #{name} for: #{args.inspect}"
-            self.user_editable_fields += args.collect(&:to_sym)
+          # Builds this job instance from the supplied properties hash.
+          # Overridden by batch to support child objects.
+          def from_properties(properties)
+            new(properties)
           end
         end
 
-        # Returns [true|false] whether to collect nil results from running this batch
-        def collect_nil_output?
-          collect_output? ? (collect_nil_output == true) : false
-        end
-
-        # Returns [true|false] whether to collect the results from running this batch
-        def collect_output?
-          collect_output == true
-        end
-
         # Returns [Float] the number of seconds the job has taken
         # - Elapsed seconds to process the job from when a worker first started working on it
         #   until now if still running, or until it was completed
@@ -282,7 +257,6 @@ module RocketJob
         # Returns [Hash] status of this job
         def as_json
           attrs = serializable_hash(methods: %i[seconds duration])
-          attrs.delete("result") unless collect_output?
           attrs.delete("failure_count") unless failure_count.positive?
           if queued?
             attrs.delete("started_at")
@@ -319,16 +293,17 @@ module RocketJob
           h = as_json
           h.delete("seconds")
           h.dup.each_pair do |k, v|
-            if v.is_a?(Time)
+            case v
+            when Time
               h[k] = v.in_time_zone(time_zone).to_s
-            elsif v.is_a?(BSON::ObjectId)
+            when BSON::ObjectId
               h[k] = v.to_s
             end
           end
           h
         end
 
-        # Returns [Boolean] whether the worker runs on a particular server.
+        # Returns [true|false] whether the worker runs on a particular server.
         def worker_on_server?(server_name)
           return false unless worker_name.present? && server_name.present?
 

data/lib/rocket_job/plugins/job/throttle.rb

@@ -48,7 +48,7 @@ module RocketJob
           # Note: Throttles are executed in the order they are defined.
           def define_throttle(method_name, filter: :throttle_filter_class)
             # Duplicate to prevent modifying parent class throttles
-            definitions = rocket_job_throttles ? rocket_job_throttles.dup : ThrottleDefinitions.new
+            definitions = rocket_job_throttles ? rocket_job_throttles.deep_dup : ThrottleDefinitions.new
             definitions.add(method_name, filter)
             self.rocket_job_throttles = definitions
           end
@@ -57,7 +57,7 @@ module RocketJob
           def undefine_throttle(method_name)
             return unless rocket_job_throttles
 
-            definitions = rocket_job_throttles.dup
+            definitions = rocket_job_throttles.deep_dup
             definitions.remove(method_name)
             self.rocket_job_throttles = definitions
           end

data/lib/rocket_job/plugins/job/throttle_running_jobs.rb

@@ -29,20 +29,28 @@ module RocketJob
           class_attribute :throttle_running_jobs
           self.throttle_running_jobs = nil
 
+          # Allow jobs to be throttled by group name instance of the job class name.
+          field :throttle_group, type: String, class_attribute: true, user_editable: true, copy_on_restart: true
+
           define_throttle :throttle_running_jobs_exceeded?
         end
 
         private
 
-        # Returns [Boolean] whether the throttle for this job has been exceeded
+        # Returns [true|false] whether the throttle for this job has been exceeded
         def throttle_running_jobs_exceeded?
-          return unless throttle_running_jobs&.positive?
+          return false unless throttle_running_jobs&.positive?
 
-          # Cannot use this class since it will include instances of parent job classes.
           RocketJob::Job.with(read: {mode: :primary}) do |conn|
-            conn.running.where("_type" => self.class.name, :id.ne => id).count >= throttle_running_jobs
+            query = throttle_running_jobs_base_query
+            throttle_group ? query["throttle_group"] = throttle_group : query["_type"] = self.class.name
+            conn.running.where(query).count >= throttle_running_jobs
           end
         end
+
+        def throttle_running_jobs_base_query
+          {:id.ne => id}
+        end
       end
     end
   end

data/lib/rocket_job/plugins/job/worker.rb

@@ -48,11 +48,11 @@ module RocketJob
         def perform_now
           raise(::Mongoid::Errors::Validations, self) unless valid?
 
-          worker = RocketJob::Worker.new(inline: true)
+          worker = RocketJob::Worker.new
           start if may_start?
           # Re-Raise exceptions
           rocket_job_work(worker, true) if running?
-          result
+          @rocket_job_output
         end
 
         def perform(*)
@@ -106,11 +106,6 @@ module RocketJob
               end
             end
 
-            if collect_output?
-              # Result must be a Hash, if not put it in a Hash
-              self.result = @rocket_job_output.is_a?(Hash) ? @rocket_job_output : {"result" => @rocket_job_output}
-            end
-
             if new_record? || destroyed?
               complete if may_complete?
             else

data/lib/rocket_job/plugins/restart.rb

@@ -91,8 +91,16 @@ module RocketJob
           logger.info("Job has expired. Not creating a new instance.")
           return
         end
-        attributes = rocket_job_restart_attributes.each_with_object({}) { |attr, attrs| attrs[attr] = send(attr) }
-        rocket_job_restart_create(attributes)
+        job_attrs             =
+          rocket_job_restart_attributes.each_with_object({}) { |attr, attrs| attrs[attr] = send(attr) }
+        job                   = self.class.new(job_attrs)
+
+        # Copy across input and output categories to new scheduled job so that all of the
+        # settings are remembered between instance. Example: slice_size
+        job.input_categories  = input_categories if respond_to?(:input_categories)
+        job.output_categories = output_categories if respond_to?(:output_categories)
+
+        rocket_job_restart_save(job)
       end
 
       def rocket_job_restart_abort
@@ -101,11 +109,10 @@ module RocketJob
 
       # Allow Singleton to prevent the creation of a new job if one is already running
       # Retry since the delete may not have persisted to disk yet.
-      def rocket_job_restart_create(attrs, retry_limit = 3, sleep_interval = 0.1)
+      def rocket_job_restart_save(job, retry_limit = 10, sleep_interval = 0.5)
         count = 0
         while count < retry_limit
-          job = self.class.create(attrs)
-          if job.persisted?
+          if job.save
             logger.info("Created a new job instance: #{job.id}")
             return true
           else

data/lib/rocket_job/plugins/state_machine.rb

@@ -51,7 +51,8 @@ module RocketJob
               # Validate methods are any of Symbol String Proc
               methods.each do |method|
                 unless method.is_a?(Symbol) || method.is_a?(String)
-                  raise(ArgumentError, "#{action}_#{event_name} currently does not support any options. Only Symbol and String method names can be supplied.")
+                  raise(ArgumentError,
+                        "#{action}_#{event_name} currently does not support any options. Only Symbol and String method names can be supplied.")
                 end
               end
               methods

data/lib/rocket_job/plugins/throttle_dependent_jobs.rb

@@ -0,0 +1,38 @@
+require "active_support/concern"
+module RocketJob
+  module Plugins
+    # Prevent this job from starting, or a batch slice from starting if the dependent jobs are running.
+    #
+    # Features:
+    # - Ensures dependent jobs won't run
+    # When the throttle has been exceeded all jobs of this class will be ignored until the
+    # next refresh. `RocketJob::Config::re_check_seconds` which by default is 60 seconds.
+    module ThrottleDependentJobs
+      extend ActiveSupport::Concern
+
+      included do
+        class_attribute :dependent_jobs
+        self.dependent_jobs = nil
+
+        define_throttle :dependent_job_exists?
+        define_batch_throttle :dependent_job_exists? if respond_to?(:define_batch_throttle)
+      end
+
+      private
+
+      # Checks if there are any dependent jobs are running
+      def dependent_job_exists?
+        return false if dependent_jobs.blank?
+
+        jobs_count = RocketJob::Job.running.where(:_type.in => dependent_jobs).count
+        return false if jobs_count.zero?
+
+        logger.info(
+          message: "#{jobs_count} Dependent Jobs are running from #{dependent_jobs.join(', ')}",
+          metric:  "#{self.class.name}/dependent_jobs_throttle"
+        )
+        true
+      end
+    end
+  end
+end

data/lib/rocket_job/ractor_worker.rb

@@ -0,0 +1,42 @@
+module RocketJob
+  # Run each worker in its own "Ractor".
+  class RactorWorker < Worker
+    attr_reader :thread
+
+    def initialize(id:, server_name:)
+      super(id: id, server_name: server_name)
+      @shutdown = Concurrent::Event.new
+      @thread   = Ractor.new(name: "rocketjob-#{id}") { run }
+    end
+
+    def alive?
+      @thread.alive?
+    end
+
+    def backtrace
+      @thread.backtrace
+    end
+
+    def join(*args)
+      @thread.join(*args)
+    end
+
+    # Send each active worker the RocketJob::ShutdownException so that stops processing immediately.
+    def kill
+      @thread.raise(Shutdown, "Shutdown due to kill request for worker: #{name}") if @thread.alive?
+    end
+
+    def shutdown?
+      @shutdown.set?
+    end
+
+    def shutdown!
+      @shutdown.set
+    end
+
+    # Returns [true|false] whether the shutdown indicator was set
+    def wait_for_shutdown?(timeout = nil)
+      @shutdown.wait(timeout)
+    end
+  end
+end

data/lib/rocket_job/server/model.rb

@@ -28,7 +28,7 @@ module RocketJob
 
         # Current state
         #   Internal use only. Do not set this field directly
-        field :state, type: Symbol, default: :starting
+        field :state, type: Mongoid::StringifiedSymbol, default: :starting
 
         index({name: 1}, background: true, unique: true)
 

data/lib/rocket_job/sliced.rb

@@ -14,78 +14,23 @@ module RocketJob
       autoload :Output, "rocket_job/sliced/writer/output"
     end
 
-    # Returns [RocketJob::Sliced::Slices] for the relevant type and category.
-    #
-    # Supports compress and encrypt with [true|false|Hash] values.
-    # When [Hash] they must specify whether the apply to the input or output collection types.
-    #
-    # Example, compress both input and output collections:
-    #   class MyJob < RocketJob::Job
-    #     include RocketJob::Batch
-    #     self.compress = true
-    #   end
-    #
-    # Example, compress just the output collections:
-    #   class MyJob < RocketJob::Job
-    #     include RocketJob::Batch
-    #     self.compress = {output: true}
-    #   end
-    #
-    # To use the specialized BZip output compressor, and the regular compressor for the input collections:
-    #   class MyJob < RocketJob::Job
-    #     include RocketJob::Batch
-    #     self.compress = {output: :bzip2, input: true}
-    #   end
-    def self.factory(type, category, job)
-      raise(ArgumentError, "Unknown type: #{type.inspect}") unless %i[input output].include?(type)
-
-      collection_name = "rocket_job.#{type}s.#{job.id}"
-      collection_name << ".#{category}" unless category == :main
-
-      args               = {collection_name: collection_name, slice_size: job.slice_size}
-      klass              = slice_class(type, job)
-      args[:slice_class] = klass if klass
-
-      if type == :input
-        RocketJob::Sliced::Input.new(args)
+    # Returns [RocketJob::Sliced::Slices] for the relevant direction and category.
+    def self.factory(direction, category, job)
+      collection_name = "rocket_job.#{direction}s.#{job.id}"
+      collection_name << ".#{category.name}" unless category.name == :main
+
+      case direction
+      when :input
+        RocketJob::Sliced::Input.new(
+          collection_name: collection_name,
+          slice_class:     category.serializer_class,
+          slice_size:      category.slice_size
+        )
+      when :output
+        RocketJob::Sliced::Output.new(collection_name: collection_name, slice_class: category.serializer_class)
       else
-        RocketJob::Sliced::Output.new(args)
-      end
-    end
-
-    private
-
-    # Parses the encrypt and compress options to determine which slice serializer to use.
-    # `encrypt` takes priority over any `compress` option.
-    def self.slice_class(type, job)
-      encrypt  = extract_value(type, job.encrypt)
-      compress = extract_value(type, job.compress)
-
-      if encrypt
-        case encrypt
-        when true
-          EncryptedSlice
-        else
-          raise(ArgumentError, "Unknown job `encrypt` value: #{compress}") unless compress.is_a?(Slices)
-          # Returns the supplied class to use for encryption.
-          encrypt
-        end
-      elsif compress
-        case compress
-        when true
-          CompressedSlice
-        when :bzip2
-          BZip2OutputSlice
-        else
-          raise(ArgumentError, "Unknown job `compress` value: #{compress}") unless compress.is_a?(Slices)
-          # Returns the supplied class to use for compression.
-          compress
-        end
+        raise(ArgumentError, "Unknown direction: #{direction.inspect}")
       end
     end
-
-    def self.extract_value(type, value)
-      value.is_a?(Hash) ? value[type] : value
-    end
   end
 end