sidekiq 7.2.3 → 7.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e20a9134fa2b226bd69fb52fef9e831efa3c598e95f17f1edcd7c5c765fdee0d
-  data.tar.gz: d9b6a8e217c753b67d7f04ef9656ac81aaa767c92818c7b3ff0d6078901f7151
+  metadata.gz: 34279eae9b8159c48dd31c7b088341136d8b785e7b853389e29e179443af1577
+  data.tar.gz: 520c6f705995df5c8e50e4e607956bcd8fe78d332e9a33fdf7a4ff25b7534f60
 SHA512:
-  metadata.gz: e88f678b545310eada86b4df6eee4b3b2a479ae59743faaf73ddeb3455738f5be80a9c5b3a3a8c4abbe7ad775c1bd928be6dca518cce4cea8e7a3a5aa6c9a199
-  data.tar.gz: 46ae598a14f1c46e5f0baabfd5a6a2f16516028340ff21a1a75e90f4e0dab8d366db9bef563f90ea15351794bd5d93f03baa74e62bfd18c42eca5594002f6b7d
+  metadata.gz: 939c535c352b1c9390e7675d75cfc3728685282187c0edd8475373c6bad1ee67391477a46e26ea3de06724e8cf2570ef759862eb37234c1f9d53f79adafd5947
+  data.tar.gz: 1a127bf11d514789f6ec808f964d47e9d459687a5b2d91ebe468d6a59a28c5f9dbf01203e1aee583311c776ca9caa41831d74a15557adf5bb0bd71cce27cffdd

data/Changes.md

@@ -2,6 +2,47 @@
 
 [Sidekiq Changes](https://github.com/sidekiq/sidekiq/blob/main/Changes.md) | [Sidekiq Pro Changes](https://github.com/sidekiq/sidekiq/blob/main/Pro-Changes.md) | [Sidekiq Enterprise Changes](https://github.com/sidekiq/sidekiq/blob/main/Ent-Changes.md)
 
+7.3.0
+----------
+
+- **NEW FEATURE** Add `Sidekiq::IterableJob`, iteration support for long-running jobs. [#6286, fatkodima]
+  Iterable jobs are interruptible and can restart quickly if
+  running during a deploy. You must ensure that `each_iteration`
+  doesn't take more than Sidekiq's `-t` timeout (default: 25 seconds). Iterable jobs must not implement `perform`.
+```ruby
+class ProcessArrayJob
+  include Sidekiq::IterableJob
+  def build_enumerator(*args, **kwargs)
+    array_enumerator(args, **kwargs)
+  end
+  def each_iteration(arg)
+    puts arg
+  end
+end
+ProcessArrayJob.perform_async(1, 2, 3)
+```
+See the [Iteration](//github.com/sidekiq/sidekiq/wiki/Iteration) wiki page and the RDoc in `Sidekiq::IterableJob`.
+This feature should be considered BETA until the next minor release.
+- **SECURITY** The Web UI no longer allows extensions to use `<script>`.
+  Adjust CSP to disallow inline scripts within the Web UI. Please see
+  `examples/webui-ext` for how to register Web UI extensions and use
+  dynamic CSS and JS. This will make Sidekiq immune to XSS attacks. [#6270]
+- Add config option, `:skip_default_job_logging` to disable Sidekiq's default
+  start/finish job logging. [#6200]
+- Allow `Sidekiq::Limiter.redis` to use Redis Cluster [#6288]
+- Retain CurrentAttributeѕ after inline execution [#6307]
+- Ignore non-existent CurrentAttributes attributes when restoring [#6341]
+- Raise default Redis {read,write,connect} timeouts from 1 to 3 seconds
+  to minimize ReadTimeoutErrors [#6162]
+- Add `logger` as a dependency since it will become bundled in Ruby 3.5 [#6320]
+- Ignore unsupported locales in the Web UI [#6313]
+
+7.2.4
+----------
+
+- Fix XSS in metrics filtering introduced in 7.2.0, CVE-2024-32887
+  Thanks to @UmerAdeemCheema for the security report.
+
 7.2.3
 ----------
 

data/README.md

@@ -86,7 +86,7 @@ Useful resources:
 * Occasional announcements are made to the [@sidekiq](https://ruby.social/@sidekiq) Mastodon account.
 * The [Sidekiq tag](https://stackoverflow.com/questions/tagged/sidekiq) on Stack Overflow has lots of useful Q & A.
 
-Every Friday morning is Sidekiq office hour: I video chat and answer questions.
+Every Thursday morning is Sidekiq office hour: I video chat and answer questions.
 See the [Sidekiq support page](https://sidekiq.org/support.html) for details.
 
 Contributing

data/lib/sidekiq/api.rb

@@ -1199,7 +1199,7 @@ module Sidekiq
       @hsh.send(*all)
     end
 
-    def respond_to_missing?(name)
+    def respond_to_missing?(name, *args)
       @hsh.respond_to?(name)
     end
   end

data/lib/sidekiq/capsule.rb

@@ -17,6 +17,7 @@ module Sidekiq
   # end
   class Capsule
     include Sidekiq::Component
+    extend Forwardable
 
     attr_reader :name
     attr_reader :queues
@@ -24,6 +25,8 @@ module Sidekiq
     attr_reader :mode
     attr_reader :weights
 
+    def_delegators :@config, :[], :[]=, :fetch, :key?, :has_key?, :merge!, :dig
+
     def initialize(name, config)
       @name = name
       @config = config

data/lib/sidekiq/cli.rb

@@ -423,3 +423,4 @@ end
 
 require "sidekiq/systemd"
 require "sidekiq/metrics/tracking"
+require "sidekiq/job/interrupt_handler"

data/lib/sidekiq/client.rb

@@ -248,10 +248,10 @@ module Sidekiq
     def atomic_push(conn, payloads)
       if payloads.first.key?("at")
         conn.zadd("schedule", payloads.flat_map { |hash|
-          at = hash.delete("at").to_s
+          at = hash["at"].to_s
           # ActiveJob sets this but the job has not been enqueued yet
           hash.delete("enqueued_at")
-          [at, Sidekiq.dump_json(hash)]
+          [at, Sidekiq.dump_json(hash.except("at"))]
         })
       else
         queue = payloads.first["queue"]

data/lib/sidekiq/config.rb

@@ -17,6 +17,10 @@ module Sidekiq
       poll_interval_average: nil,
       average_scheduled_poll_interval: 5,
       on_complex_arguments: :raise,
+      iteration: {
+        max_job_runtime: nil,
+        retry_backoff: 0
+      },
       error_handlers: [],
       death_handlers: [],
       lifecycle_events: {
@@ -52,7 +56,7 @@ module Sidekiq
       @capsules = {}
     end
 
-    def_delegators :@options, :[], :[]=, :fetch, :key?, :has_key?, :merge!
+    def_delegators :@options, :[], :[]=, :fetch, :key?, :has_key?, :merge!, :dig
     attr_reader :capsules
 
     def to_json(*)

data/lib/sidekiq/fetch.rb

@@ -44,7 +44,7 @@ module Sidekiq # :nodoc:
         return nil
       end
 
-      queue, job = redis { |conn| conn.blocking_call(conn.read_timeout + TIMEOUT, "brpop", *qs, TIMEOUT) }
+      queue, job = redis { |conn| conn.blocking_call(TIMEOUT, "brpop", *qs, TIMEOUT) }
       UnitOfWork.new(queue, job, config) if queue
     end
 

data/lib/sidekiq/iterable_job.rb

@@ -0,0 +1,53 @@
+require "sidekiq/job/iterable"
+
+# Iterable jobs are ones which provide a sequence to process using
+# `build_enumerator(*args, cursor: cursor)` and then process each
+# element of that sequence in `each_iteration(item, *args)`.
+#
+# The job is kicked off as normal:
+#
+#     ProcessUserSet.perform_async(123)
+#
+# but instead of calling `perform`, Sidekiq will call:
+#
+#     enum = ProcessUserSet#build_enumerator(123, cursor:nil)
+#
+# Your Enumerator must yield `(object, updated_cursor)` and
+# Sidekiq will call your `each_iteration` method:
+#
+#     ProcessUserSet#each_iteration(object, 123)
+#
+# After every iteration, Sidekiq will check for shutdown. If we are
+# stopping, the cursor will be saved to Redis and the job re-queued
+# to pick up the rest of the work upon restart. Your job will get
+# the updated_cursor so it can pick up right where it stopped.
+#
+#     enum = ProcessUserSet#build_enumerator(123, cursor: updated_cursor)
+#
+# The cursor object must be serializable to JSON.
+#
+# Note there are several APIs to help you build enumerators for
+# ActiveRecord Relations, CSV files, etc. See sidekiq/job/iterable/*.rb.
+module Sidekiq
+  module IterableJob
+    def self.included(base)
+      base.include Sidekiq::Job
+      base.include Sidekiq::Job::Iterable
+    end
+
+    # def build_enumerator(*args, cursor:)
+    # def each_iteration(item, *args)
+
+    # Your job can also define several callbacks during points
+    # in each job's lifecycle.
+    #
+    # def on_start
+    # def on_resume
+    # def on_stop
+    # def on_complete
+    # def around_iteration
+    #
+    # To keep things simple and compatible, this is the same
+    # API as the `sidekiq-iteration` gem.
+  end
+end

data/lib/sidekiq/job/interrupt_handler.rb

@@ -0,0 +1,22 @@
+module Sidekiq
+  module Job
+    class InterruptHandler
+      include Sidekiq::ServerMiddleware
+
+      def call(instance, hash, queue)
+        yield
+      rescue Interrupted
+        logger.debug "Interrupted, re-queueing..."
+        c = Sidekiq::Client.new
+        c.push(hash)
+        raise Sidekiq::JobRetry::Skip
+      end
+    end
+  end
+end
+
+Sidekiq.configure_server do |config|
+  config.server_middleware do |chain|
+    chain.add Sidekiq::Job::InterruptHandler
+  end
+end

data/lib/sidekiq/job/iterable/active_record_enumerator.rb

@@ -0,0 +1,53 @@
+# frozen_string_literal: true
+
+module Sidekiq
+  module Job
+    module Iterable
+      # @api private
+      class ActiveRecordEnumerator
+        def initialize(relation, cursor: nil, **options)
+          @relation = relation
+          @cursor = cursor
+          @options = options
+        end
+
+        def records
+          Enumerator.new(-> { @relation.count }) do |yielder|
+            @relation.find_each(**@options, start: @cursor) do |record|
+              yielder.yield(record, record.id)
+            end
+          end
+        end
+
+        def batches
+          Enumerator.new(-> { @relation.count }) do |yielder|
+            @relation.find_in_batches(**@options, start: @cursor) do |batch|
+              yielder.yield(batch, batch.last.id)
+            end
+          end
+        end
+
+        def relations
+          Enumerator.new(-> { relations_size }) do |yielder|
+            # Convenience to use :batch_size for all the
+            # ActiveRecord batching methods.
+            options = @options.dup
+            options[:of] ||= options.delete(:batch_size)
+
+            @relation.in_batches(**options, start: @cursor) do |relation|
+              last_record = relation.last
+              yielder.yield(relation, last_record.id)
+            end
+          end
+        end
+
+        private
+
+        def relations_size
+          batch_size = @options[:batch_size] || 1000
+          (@relation.count + batch_size - 1) / batch_size # ceiling division
+        end
+      end
+    end
+  end
+end

data/lib/sidekiq/job/iterable/csv_enumerator.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module Sidekiq
+  module Job
+    module Iterable
+      # @api private
+      class CsvEnumerator
+        def initialize(csv)
+          unless defined?(CSV) && csv.instance_of?(CSV)
+            raise ArgumentError, "CsvEnumerator.new takes CSV object"
+          end
+
+          @csv = csv
+        end
+
+        def rows(cursor:)
+          @csv.lazy
+            .each_with_index
+            .drop(cursor || 0)
+            .to_enum { count_of_rows_in_file }
+        end
+
+        def batches(cursor:, batch_size: 100)
+          @csv.lazy
+            .each_slice(batch_size)
+            .with_index
+            .drop(cursor || 0)
+            .to_enum { (count_of_rows_in_file.to_f / batch_size).ceil }
+        end
+
+        private
+
+        def count_of_rows_in_file
+          filepath = @csv.path
+          return unless filepath
+
+          count = IO.popen(["wc", "-l", filepath]) do |out|
+            out.read.strip.to_i
+          end
+
+          count -= 1 if @csv.headers
+          count
+        end
+      end
+    end
+  end
+end

data/lib/sidekiq/job/iterable/enumerators.rb

@@ -0,0 +1,135 @@
+# frozen_string_literal: true
+
+require_relative "active_record_enumerator"
+require_relative "csv_enumerator"
+
+module Sidekiq
+  module Job
+    module Iterable
+      module Enumerators
+        # Builds Enumerator object from a given array, using +cursor+ as an offset.
+        #
+        # @param array [Array]
+        # @param cursor [Integer] offset to start iteration from
+        #
+        # @return [Enumerator]
+        #
+        # @example
+        #   array_enumerator(['build', 'enumerator', 'from', 'any', 'array'], cursor: cursor)
+        #
+        def array_enumerator(array, cursor:)
+          raise ArgumentError, "array must be an Array" unless array.is_a?(Array)
+
+          x = array.each_with_index.drop(cursor || 0)
+          x.to_enum { x.size }
+        end
+
+        # Builds Enumerator from `ActiveRecord::Relation`.
+        # Each Enumerator tick moves the cursor one row forward.
+        #
+        # @param relation [ActiveRecord::Relation] relation to iterate
+        # @param cursor [Object] offset id to start iteration from
+        # @param options [Hash] additional options that will be passed to relevant
+        #   ActiveRecord batching methods
+        #
+        # @return [ActiveRecordEnumerator]
+        #
+        # @example
+        #   def build_enumerator(cursor:)
+        #     active_record_records_enumerator(User.all, cursor: cursor)
+        #   end
+        #
+        #   def each_iteration(user)
+        #     user.notify_about_something
+        #   end
+        #
+        def active_record_records_enumerator(relation, cursor:, **options)
+          ActiveRecordEnumerator.new(relation, cursor: cursor, **options).records
+        end
+
+        # Builds Enumerator from `ActiveRecord::Relation` and enumerates on batches of records.
+        # Each Enumerator tick moves the cursor `:batch_size` rows forward.
+        # @see #active_record_records_enumerator
+        #
+        # @example
+        #   def build_enumerator(product_id, cursor:)
+        #     active_record_batches_enumerator(
+        #       Comment.where(product_id: product_id).select(:id),
+        #       cursor: cursor,
+        #       batch_size: 100
+        #     )
+        #   end
+        #
+        #   def each_iteration(batch_of_comments, product_id)
+        #     comment_ids = batch_of_comments.map(&:id)
+        #     CommentService.call(comment_ids: comment_ids)
+        #   end
+        #
+        def active_record_batches_enumerator(relation, cursor:, **options)
+          ActiveRecordEnumerator.new(relation, cursor: cursor, **options).batches
+        end
+
+        # Builds Enumerator from `ActiveRecord::Relation` and enumerates on batches,
+        # yielding `ActiveRecord::Relation`s.
+        # @see #active_record_records_enumerator
+        #
+        # @example
+        #   def build_enumerator(product_id, cursor:)
+        #     active_record_relations_enumerator(
+        #       Product.find(product_id).comments,
+        #       cursor: cursor,
+        #       batch_size: 100,
+        #     )
+        #   end
+        #
+        #   def each_iteration(batch_of_comments, product_id)
+        #     # batch_of_comments will be a Comment::ActiveRecord_Relation
+        #     batch_of_comments.update_all(deleted: true)
+        #   end
+        #
+        def active_record_relations_enumerator(relation, cursor:, **options)
+          ActiveRecordEnumerator.new(relation, cursor: cursor, **options).relations
+        end
+
+        # Builds Enumerator from a CSV file.
+        #
+        # @param csv [CSV] an instance of CSV object
+        # @param cursor [Integer] offset to start iteration from
+        #
+        # @example
+        #   def build_enumerator(import_id, cursor:)
+        #     import = Import.find(import_id)
+        #     csv_enumerator(import.csv, cursor: cursor)
+        #   end
+        #
+        #   def each_iteration(csv_row)
+        #     # insert csv_row into database
+        #   end
+        #
+        def csv_enumerator(csv, cursor:)
+          CsvEnumerator.new(csv).rows(cursor: cursor)
+        end
+
+        # Builds Enumerator from a CSV file and enumerates on batches of records.
+        #
+        # @param csv [CSV] an instance of CSV object
+        # @param cursor [Integer] offset to start iteration from
+        # @option options :batch_size [Integer] (100) size of the batch
+        #
+        # @example
+        #   def build_enumerator(import_id, cursor:)
+        #     import = Import.find(import_id)
+        #     csv_batches_enumerator(import.csv, cursor: cursor)
+        #   end
+        #
+        #   def each_iteration(batch_of_csv_rows)
+        #     # ...
+        #   end
+        #
+        def csv_batches_enumerator(csv, cursor:, **options)
+          CsvEnumerator.new(csv).batches(cursor: cursor, **options)
+        end
+      end
+    end
+  end
+end

data/lib/sidekiq/job/iterable.rb

@@ -0,0 +1,231 @@
+# frozen_string_literal: true
+
+require_relative "iterable/enumerators"
+
+module Sidekiq
+  module Job
+    class Interrupted < ::RuntimeError; end
+
+    module Iterable
+      include Enumerators
+
+      # @api private
+      def self.included(base)
+        base.extend(ClassMethods)
+      end
+
+      # @api private
+      module ClassMethods
+        def method_added(method_name)
+          raise "#{self} is an iterable job and must not define #perform" if method_name == :perform
+          super
+        end
+      end
+
+      # @api private
+      def initialize
+        super
+
+        @_executions = 0
+        @_cursor = nil
+        @_start_time = nil
+        @_runtime = 0
+      end
+
+      # A hook to override that will be called when the job starts iterating.
+      #
+      # It is called only once, for the first time.
+      #
+      def on_start
+      end
+
+      # A hook to override that will be called around each iteration.
+      #
+      # Can be useful for some metrics collection, performance tracking etc.
+      #
+      def around_iteration
+        yield
+      end
+
+      # A hook to override that will be called when the job resumes iterating.
+      #
+      def on_resume
+      end
+
+      # A hook to override that will be called each time the job is interrupted.
+      #
+      # This can be due to interruption or sidekiq stopping.
+      #
+      def on_stop
+      end
+
+      # A hook to override that will be called when the job finished iterating.
+      #
+      def on_complete
+      end
+
+      # The enumerator to be iterated over.
+      #
+      # @return [Enumerator]
+      #
+      # @raise [NotImplementedError] with a message advising subclasses to
+      #     implement an override for this method.
+      #
+      def build_enumerator(*)
+        raise NotImplementedError, "#{self.class.name} must implement a '#build_enumerator' method"
+      end
+
+      # The action to be performed on each item from the enumerator.
+      #
+      # @return [void]
+      #
+      # @raise [NotImplementedError] with a message advising subclasses to
+      #     implement an override for this method.
+      #
+      def each_iteration(*)
+        raise NotImplementedError, "#{self.class.name} must implement an '#each_iteration' method"
+      end
+
+      def iteration_key
+        "it-#{jid}"
+      end
+
+      # @api private
+      def perform(*arguments)
+        fetch_previous_iteration_state
+
+        @_executions += 1
+        @_start_time = ::Process.clock_gettime(::Process::CLOCK_MONOTONIC)
+
+        enumerator = build_enumerator(*arguments, cursor: @_cursor)
+        unless enumerator
+          logger.info("'#build_enumerator' returned nil, skipping the job.")
+          return
+        end
+
+        assert_enumerator!(enumerator)
+
+        if @_executions == 1
+          on_start
+        else
+          on_resume
+        end
+
+        completed = catch(:abort) do
+          iterate_with_enumerator(enumerator, arguments)
+        end
+
+        on_stop
+        completed = handle_completed(completed)
+
+        if completed
+          on_complete
+          cleanup
+        else
+          reenqueue_iteration_job
+        end
+      end
+
+      private
+
+      def fetch_previous_iteration_state
+        state = Sidekiq.redis { |conn| conn.hgetall(iteration_key) }
+
+        unless state.empty?
+          @_executions = state["ex"].to_i
+          @_cursor = Sidekiq.load_json(state["c"])
+          @_runtime = state["rt"].to_f
+        end
+      end
+
+      STATE_FLUSH_INTERVAL = 5 # seconds
+      # we need to keep the state around as long as the job
+      # might be retrying
+      STATE_TTL = 30 * 24 * 60 * 60 # one month
+
+      def iterate_with_enumerator(enumerator, arguments)
+        found_record = false
+        state_flushed_at = ::Process.clock_gettime(::Process::CLOCK_MONOTONIC)
+
+        enumerator.each do |object, cursor|
+          found_record = true
+          @_cursor = cursor
+
+          is_interrupted = interrupted?
+          if ::Process.clock_gettime(::Process::CLOCK_MONOTONIC) - state_flushed_at >= STATE_FLUSH_INTERVAL || is_interrupted
+            flush_state
+            state_flushed_at = ::Process.clock_gettime(::Process::CLOCK_MONOTONIC)
+          end
+
+          return false if is_interrupted
+
+          around_iteration do
+            each_iteration(object, *arguments)
+          end
+        end
+
+        logger.debug("Enumerator found nothing to iterate!") unless found_record
+        true
+      ensure
+        @_runtime += (::Process.clock_gettime(::Process::CLOCK_MONOTONIC) - @_start_time)
+      end
+
+      def reenqueue_iteration_job
+        flush_state
+        logger.debug { "Interrupting job (cursor=#{@_cursor.inspect})" }
+
+        raise Interrupted
+      end
+
+      def assert_enumerator!(enum)
+        unless enum.is_a?(Enumerator)
+          raise ArgumentError, <<~MSG
+            #build_enumerator must return an Enumerator, but returned #{enum.class}.
+            Example:
+              def build_enumerator(params, cursor:)
+                active_record_records_enumerator(
+                  Shop.find(params["shop_id"]).products,
+                  cursor: cursor
+                )
+              end
+          MSG
+        end
+      end
+
+      def flush_state
+        key = iteration_key
+        state = {
+          "ex" => @_executions,
+          "c" => Sidekiq.dump_json(@_cursor),
+          "rt" => @_runtime
+        }
+
+        Sidekiq.redis do |conn|
+          conn.multi do |pipe|
+            pipe.hset(key, state)
+            pipe.expire(key, STATE_TTL)
+          end
+        end
+      end
+
+      def cleanup
+        logger.debug {
+          format("Completed iteration. executions=%d runtime=%.3f", @_executions, @_runtime)
+        }
+        Sidekiq.redis { |conn| conn.unlink(iteration_key) }
+      end
+
+      def handle_completed(completed)
+        case completed
+        when nil, # someone aborted the job but wants to call the on_complete callback
+             true
+          true
+        when false
+          false
+        else
+          raise "Unexpected thrown value: #{completed.inspect}"
+        end
+      end
+    end
+  end
+end