karafka-rdkafka 0.16.0 → 0.17.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 78fe68d4a566b63ffa94633d2752cc95c535c020236749aa0452192d10cd2c43
-  data.tar.gz: 61a9fe426703153762c94508a535a780b9ee1917cacf93ccbe37d1c4eb974d6c
+  metadata.gz: f0a60c074ecbf580c271442ac5e86a2152acf2542fc88c841f3adf8de99a8326
+  data.tar.gz: ab81215e4a2d4d641521efbd8f0e679a169bea6920378fab237f8401b0b8f3ac
 SHA512:
-  metadata.gz: 46e2862b087c375042b93682d10378901d23b238f696cfd17048a880c3b28170243250e5cceecfd42c74c98791273df9577ed1371b300629a60727df8df2f9c4
-  data.tar.gz: c76d0666e617aa0879ffa99c1888b7fad0ebbd1f1195f14285b130b82ba34c339c3f92c9718f0e8180d2d1cccd1e99f23ea765ea314fb0570713ce4ba23b8071
+  metadata.gz: 9595e54c497e0018da1df9611d1b2dca8b4e03fab8ca5ef26e3e20da93fb2ccdc9c0f568ddf7c371fe4f59873187e653744f087afea1dc6ebfc57a151b572dfc
+  data.tar.gz: 6ce0c6d3baee8e939dcfe1867771d2ca41b8d8c9d6018a5d00e63e5c9d79c499beebfdca3f75d8624862351191e36d4ec20fb86fad3eaeb8f169dca2802e83c2

data/.ruby-version

@@ -1 +1 @@
-3.3.3
+3.3.4

data/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Rdkafka Changelog
 
+## 0.17.0 (2024-07-21)
+- [Enhancement] Bump librdkafka to 2.5.0
+
+## 0.16.1 (2024-07-10)
+- [Feature] Add `#seek_by` to be able to seek for a message by topic, partition and offset (zinahia)
+- [Fix] Switch to local release of librdkafka to mitigate its unavailability.
+
 ## 0.16.0 (2024-06-17)
 - **[Breaking]** Messages without headers returned by `#poll` contain frozen empty hash.
 - **[Breaking]** `HashWithSymbolKeysTreatedLikeStrings` has been removed so headers are regular hashes with string keys.
@@ -7,6 +14,9 @@
 - [Enhancement] Save two objects on message produced and lower CPU usage on message produced with small improvements.
 - [Fix] Remove support for Ruby 2.7. Supporting it was a bug since rest of the karafka ecosystem no longer supports it.
 
+## 0.15.2 (2024-07-10)
+- [Fix] Switch to local release of librdkafka to mitigate its unavailability.
+
 ## 0.15.1 (2024-05-09)
 - **[Feature]** Provide ability to use topic config on a producer for custom behaviors per dispatch.
 - [Enhancement] Use topic config reference cache for messages production to prevent topic objects allocation with each message.
@@ -17,6 +27,9 @@
 - **[Feature]** Support incremental config describe + alter API (mensfeld)
 - [Enhancement] name polling Thread as `rdkafka.native_kafka#<name>` (nijikon)
 
+## 0.14.11 (2024-07-10)
+- [Fix] Switch to local release of librdkafka to mitigate its unavailability.
+
 ## 0.14.10 (2024-02-08)
 - [Fix] Background logger stops working after forking causing memory leaks (mensfeld).
 
@@ -63,6 +76,9 @@
 - [Enhancement] Bump librdkafka to 2.3.0
 - [Enhancement] Increase the `#lag` and `#query_watermark_offsets` default timeouts from 100ms to 1000ms. This will compensate for network glitches and remote clusters operations.
 
+## 0.13.10 (2024-07-10)
+- [Fix] Switch to local release of librdkafka to mitigate its unavailability.
+
 ## 0.13.9 (2023-11-07)
 - [Enhancement] Expose alternative way of managing consumer events via a separate queue.
 - [Enhancement] Allow for setting `statistics_callback` as nil to reset predefined settings configured by a different gem.
@@ -117,6 +133,9 @@
 - Retry metadta fetches on certain errors with a backoff (mensfeld)
 - Do not lock access to underlying native kafka client and rely on Karafka granular locking (mensfeld)
 
+## 0.12.4 (2024-07-10)
+- [Fix] Switch to local release of librdkafka to mitigate its unavailability.
+
 ## 0.12.3
 - Include backtrace in non-raised binded errors.
 - Include topic name in the delivery reports

data/README.md

@@ -163,7 +163,8 @@ bundle exec rake produce_messages
 
 | rdkafka-ruby | librdkafka |
 |-|-|
-| 0.16.0 (2024-06-13) | 2.3.0 (2023-10-25) |
+| 0.17.0 (Unreleased) | 2.5.0 (2024-07-10) |
+| 0.16.0 (2024-06-13) | 2.4.0 (2024-05-07) |
 | 0.15.0 (2023-12-03) | 2.3.0 (2023-10-25) |
 | 0.14.0 (2023-11-21) | 2.2.0 (2023-07-12) |
 | 0.13.0 (2023-07-24) | 2.0.2 (2023-01-20) |

data/docker-compose.yml

@@ -3,7 +3,7 @@ version: '2'
 services:
   kafka:
     container_name: kafka
-    image: confluentinc/cp-kafka:7.6.1
+    image: confluentinc/cp-kafka:7.6.2
 
     ports:
       - 9092:9092

data/ext/README.md

@@ -1,7 +1,7 @@
 # Ext
 
-This gem depends on the `librdkafka` C library. It is downloaded when
-this gem is installed.
+This gem depends on the `librdkafka` C library. It is downloaded, stored in
+`dist/` directory, and checked into source control.
 
 To update the `librdkafka` version follow the following steps:
 
@@ -9,8 +9,9 @@ To update the `librdkafka` version follow the following steps:
   version number and asset checksum for `tar.gz`.
 * Change the version in `lib/rdkafka/version.rb`
 * Change the `sha256` in `lib/rdkafka/version.rb`
-* Run `bundle exec rake` in the `ext` directory to download and build
-  the new version
+* Run `bundle exec rake dist:download` in the `ext` directory to download the
+  new release and place it in the `dist/` for you
+* Run `bundle exec rake` in the `ext` directory to build the new version
 * Run `docker-compose pull` in the main gem directory to ensure the docker
   images used by the tests and run `docker-compose up`
 * Finally, run `bundle exec rspec` in the main gem directory to execute

data/ext/Rakefile

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
 require File.expand_path('../../lib/rdkafka/version', __FILE__)
+require "digest"
 require "fileutils"
 require "open-uri"
 
@@ -22,8 +23,10 @@ task :default => :clean do
       ENV["LDFLAGS"] = "-L#{homebrew_prefix}/lib" unless ENV["LDFLAGS"]
     end
 
+    releases = File.expand_path(File.join(File.dirname(__FILE__), '../dist'))
+
     recipe.files << {
-      :url => "https://codeload.github.com/edenhill/librdkafka/tar.gz/v#{Rdkafka::LIBRDKAFKA_VERSION}",
+      :url => "file://#{releases}/librdkafka_#{Rdkafka::LIBRDKAFKA_VERSION}.tar.gz",
       :sha256 => Rdkafka::LIBRDKAFKA_SOURCE_SHA256
     }
     recipe.configure_options = ["--host=#{recipe.host}"]
@@ -69,6 +72,42 @@ task :clean do
   FileUtils.rm_rf File.join(File.dirname(__FILE__), "tmp")
 end
 
+namespace :dist do
+  task :dir do
+    ENV["RDKAFKA_DIST_PATH"] ||= File.expand_path(File.join(File.dirname(__FILE__), '..', 'dist'))
+  end
+
+  task :file => "dist:dir" do
+    ENV["RDKAFKA_DIST_FILE"] ||= File.join(ENV["RDKAFKA_DIST_PATH"], "librdkafka_#{Rdkafka::LIBRDKAFKA_VERSION}.tar.gz")
+  end
+
+  task :clean => "dist:file" do
+    Dir.glob(File.join("#{ENV['RDKAFKA_DIST_PATH']}", "*")).each do |filename|
+      next if filename.include? ENV["RDKAFKA_DIST_FILE"]
+
+      FileUtils.rm_rf filename
+    end
+  end
+
+  task :download => "dist:file" do
+    version = Rdkafka::LIBRDKAFKA_VERSION
+    librdkafka_download = "https://codeload.github.com/confluentinc/librdkafka/tar.gz/v#{version}"
+
+    URI.open(librdkafka_download) do |file|
+      filename = ENV["RDKAFKA_DIST_FILE"]
+      data     = file.read
+
+      if Digest::SHA256.hexdigest(data) != Rdkafka::LIBRDKAFKA_SOURCE_SHA256
+        raise "SHA256 does not match downloaded file"
+      end
+
+      File.write(filename, data)
+    end
+  end
+
+  task :update => %w[dist:download dist:clean]
+end
+
 namespace :build do
   desc "Build librdkafka at the given git sha or tag"
   task :git, [:ref] do |task, args|
@@ -76,7 +115,7 @@ namespace :build do
     version = "git-#{ref}"
 
     recipe = MiniPortile.new("librdkafka", version)
-    recipe.files << "https://github.com/edenhill/librdkafka/archive/#{ref}.tar.gz"
+    recipe.files << "https://github.com/confluentinc/librdkafka/archive/#{ref}.tar.gz"
     recipe.configure_options = ["--host=#{recipe.host}","--enable-static", "--enable-zstd"]
     recipe.cook
 

data/lib/rdkafka/consumer.rb

@@ -435,6 +435,19 @@ module Rdkafka
     # @return [nil]
     # @raise [RdkafkaError] When seeking fails
     def seek(message)
+      seek_by(message.topic, message.partition, message.offset)
+    end
+
+    # Seek to a particular message by providing the topic, partition and offset.
+    # The next poll on the topic/partition will return the
+    # message at the given offset.
+    #
+    # @param topic [String] The topic in which to seek
+    # @param partition [Integer] The partition number to seek
+    # @param offset [Integer] The partition offset to seek
+    # @return [nil]
+    # @raise [RdkafkaError] When seeking fails
+    def seek_by(topic, partition, offset)
       closed_consumer_check(__method__)
 
       # rd_kafka_offset_store is one of the few calls that does not support
@@ -442,14 +455,14 @@ module Rdkafka
       native_topic = @native_kafka.with_inner do |inner|
         Rdkafka::Bindings.rd_kafka_topic_new(
           inner,
-          message.topic,
+          topic,
           nil
         )
       end
       response = Rdkafka::Bindings.rd_kafka_seek(
         native_topic,
-        message.partition,
-        message.offset,
+        partition,
+        offset,
         0 # timeout
       )
       Rdkafka::RdkafkaError.validate!(response)

data/lib/rdkafka/version.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module Rdkafka
-  VERSION = "0.16.0"
-  LIBRDKAFKA_VERSION = "2.4.0"
-  LIBRDKAFKA_SOURCE_SHA256 = "d645e47d961db47f1ead29652606a502bdd2a880c85c1e060e94eea040f1a19a"
+  VERSION = "0.17.0"
+  LIBRDKAFKA_VERSION = "2.5.0"
+  LIBRDKAFKA_SOURCE_SHA256 = "3dc62de731fd516dfb1032861d9a580d4d0b5b0856beb0f185d06df8e6c26259"
 end

data/spec/rdkafka/admin_spec.rb

@@ -34,7 +34,7 @@ describe Rdkafka::Admin do
   describe '#describe_errors' do
     let(:errors) { admin.class.describe_errors }
 
-    it { expect(errors.size).to eq(168) }
+    it { expect(errors.size).to eq(170) }
     it { expect(errors[-184]).to eq(code: -184, description: 'Local: Queue full', name: '_QUEUE_FULL') }
     it { expect(errors[21]).to eq(code: 21, description: 'Broker: Invalid required acks value', name: 'INVALID_REQUIRED_ACKS') }
   end

data/spec/rdkafka/consumer_spec.rb

@@ -258,6 +258,95 @@ describe Rdkafka::Consumer do
     end
   end
 
+  describe "#seek_by" do
+    let(:topic) { "consume_test_topic" }
+    let(:partition) { 0 }
+    let(:offset) { 0 }
+
+    it "should raise an error when seeking fails" do
+      expect(Rdkafka::Bindings).to receive(:rd_kafka_seek).and_return(20)
+      expect {
+        consumer.seek_by(topic, partition, offset)
+      }.to raise_error Rdkafka::RdkafkaError
+    end
+
+    context "subscription" do
+      let(:timeout) { 1000 }
+
+      before do
+        consumer.subscribe(topic)
+
+        # 1. partitions are assigned
+        wait_for_assignment(consumer)
+        expect(consumer.assignment).not_to be_empty
+
+        # 2. eat unrelated messages
+        while(consumer.poll(timeout)) do; end
+      end
+      after { consumer.unsubscribe }
+
+      def send_one_message(val)
+        producer.produce(
+          topic:     topic,
+          payload:   "payload #{val}",
+          key:       "key 1",
+          partition: 0
+        ).wait
+      end
+
+      it "works when a partition is paused" do
+        # 3. get reference message
+        send_one_message(:a)
+        message1 = consumer.poll(timeout)
+        expect(message1&.payload).to eq "payload a"
+
+        # 4. pause the subscription
+        tpl = Rdkafka::Consumer::TopicPartitionList.new
+        tpl.add_topic(topic, 1)
+        consumer.pause(tpl)
+
+        # 5. seek by the previous message fields
+        consumer.seek_by(message1.topic, message1.partition, message1.offset)
+
+        # 6. resume the subscription
+        tpl = Rdkafka::Consumer::TopicPartitionList.new
+        tpl.add_topic(topic, 1)
+        consumer.resume(tpl)
+
+        # 7. ensure same message is read again
+        message2 = consumer.poll(timeout)
+
+        # This is needed because `enable.auto.offset.store` is true but when running in CI that
+        # is overloaded, offset store lags
+        sleep(2)
+
+        consumer.commit
+        expect(message1.offset).to eq message2.offset
+        expect(message1.payload).to eq message2.payload
+      end
+
+      it "allows skipping messages" do
+        # 3. send messages
+        send_one_message(:a)
+        send_one_message(:b)
+        send_one_message(:c)
+
+        # 4. get reference message
+        message = consumer.poll(timeout)
+        expect(message&.payload).to eq "payload a"
+
+        # 5. seek over one message
+        consumer.seek_by(message.topic, message.partition, message.offset + 2)
+
+        # 6. ensure that only one message is available
+        records = consumer.poll(timeout)
+        expect(records&.payload).to eq "payload c"
+        records = consumer.poll(timeout)
+        expect(records).to be_nil
+      end
+    end
+  end
+
   describe "#assign and #assignment" do
     it "should return an empty assignment if nothing is assigned" do
       expect(consumer.assignment).to be_empty

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: karafka-rdkafka
 version: !ruby/object:Gem::Version
-  version: 0.16.0
+  version: 0.17.0
 platform: ruby
 authors:
 - Thijs Cadier
@@ -36,7 +36,7 @@ cert_chain:
   AnG1dJU+yL2BK7vaVytLTstJME5mepSZ46qqIJXMuWob/YPDmVaBF39TDSG9e34s
   msG3BiCqgOgHAnL23+CN3Rt8MsuRfEtoTKpJVcCfoEoNHOkc
   -----END CERTIFICATE-----
-date: 2024-06-17 00:00:00.000000000 Z
+date: 2024-07-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -186,6 +186,7 @@ files:
 - README.md
 - Rakefile
 - certs/cert_chain.pem
+- dist/librdkafka_2.5.0.tar.gz
 - docker-compose.yml
 - ext/README.md
 - ext/Rakefile
@@ -285,7 +286,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.11
+rubygems_version: 3.5.9
 signing_key: 
 specification_version: 4
 summary: The rdkafka gem is a modern Kafka client library for Ruby based on librdkafka.