Codestin Search App

[EOL] Remove Rails 7.1 support according to EOL while not blocking Rails 7.1 usage.
[Enhancement] Retry on the KIP-848 stale_member_epoch error.
[Enhancement] Provide Karafka::Admin.trigger_rebalance API to programmatically trigger consumer group rebalances for operational purposes.
[Enhancement] Nest pause configuration under config.pause.* namespace (config.pause.timeout, config.pause.max_timeout, config.pause.with_exponential_backoff) while maintaining backwards compatibility with the old flat API (config.pause_timeout, etc.) via delegation methods that will be removed in Karafka 2.6.
[Enhancement] Detect and track involuntary assignment loss during long-running processing that exceeds max.poll.interval.ms via client.events_poll event and automatically update Karafka::App.assignments to reflect reality.
[Enhancement] Extend Karafka::Admin.read_watermark_offsets to accept either a single topic with partition or a hash of multiple topics with partitions, using a single consumer instance for improved efficiency when querying multiple partitions.
[Enhancement] Add configurable Karafka::ActiveJob::Deserializer to support custom serialization formats (Avro, Protobuf, etc.) for ActiveJob payloads.
[Fix] Fix ActiveJob Continuation invalid class reference bug.

[Breaking] Remove Ruby 3.1 support according to EOL.
[Feature] Support Swarm mode on MacOS.
[Enhancement] Support past dispatch_at times with jitter: 0 in the OSS Karafka to support ActiveJob continuation.
[Enhancement] Use direct topic dispatches when dispatch_at is used for past times to bypass Scheduled Messages flow.
[Enhancement] Support immediate error raising with auto.offset.reset set to error.
[Enhancement] Don't create not needed dirs in the non-Rails setup template.
[Enhancement] Improve printing of TTIN to separate threads.
[Enhancement] Support transactional ID resource in the Admin ACL management.
[Enhancement] Move post-forceful shutdown termination wait value to a config.
[Enhancement] Include consumer group, subscription group and other details in error logs for key error locations.
[Enhancement] Inherit from ActiveJob::QueueAdapters::AbstractAdapter when possible for ActiveJob base class.
[Enhancement] Disable Nagle algorithm by default for improved network performance.
[Enhancement] Optimize the messages buffer array memory allocation pattern.
[Maintenance] Add basic direct DD integration spec via DD gem karafka monitoring feature.
[Maintenance] Add integration specs for WaterDrop connection pool usage from within consumers.
[Refactoring] Comprehensive Admin module refactoring: Extract topic operations into Admin::Topics class and consumer group operations into Admin::ConsumerGroups class with proper inheritance hierarchy, cross-class method usage optimization, and constants moved to appropriate locations where they are actually used.
[Refactoring] Move routing-related contracts from Karafka::Contracts:: to Karafka::Routing::Contracts:: namespace and reorganize error message structure in YAML files under routing: scope for better code organization and logical grouping.
[Refactoring] Move config-related contracts from Karafka::Contracts::Config to Karafka::Setup::Contracts::Config namespace and reorganize error message structure in YAML files under setup: scope for better code organization and logical grouping.
[Refactoring] Move CLI server contracts from Karafka::Contracts::ServerCliOptions to Karafka::Cli::Contracts::Server namespace and reorganize error message structure in YAML files under cli: scope for improved naming consistency and logical grouping.
[Refactoring] Replace execution mode symbol-based checks with dedicated ExecutionMode class providing cleaner API with query methods (#swarm?, #embedded?) and state change methods (#swarm!, #embedded!) for improved type safety and code clarity.
[Refactoring] Replace connection client mode symbol-based checks with dedicated Connection::Mode class providing cleaner API with query methods (#subscribe?, #assign?) and state change methods (#subscribe!, #assign!) for improved code clarity.
[Fix] Improve same timestamp dispatch in scheduled messages on Ruby 3.2.
[Fix] Fix incorrect (6 seconds vs 60 seconds) reset of connections on non-recoverable errors.
[Fix] Introduce mutex-safe and thread-safe #inspect where needed.
[Fix] Fix too loose requirement of Ruby 3.0 when it was 3.1 via transitive dependencies.
[Fix] Fix Pro Cleaner Messages compatibility with external libraries that prepend modules to #each method (e.g., DataDog tracing).
[Fix] SG exclusion in swarm triggers a contract validation error.
[Change] Require waterdrop >= 2.8.10 to support new features.
[Change] Require karafka-rdkafka >= 0.22.0 to support new features and require SSL-bug free version.
[Change] Remove no longer needed cooperative.sticky rebalance patch.
[Change] Normalize how libs and deps are required (no functional change for the end user)
[Change] Remove Ruby 3.1 specs according to the EOL schedule.

[Feature] Support Swarm mode on MacOS.
[Enhancement] Support past dispatch_at times with jitter: 0 in the OSS Karafka to support ActiveJob continuation.
[Enhancement] Use direct topic dispatches when dispatch_at is used for past times to bypass Scheduled Messages flow.
[Enhancement] Support immediate error raising with auto.offset.reset set to error.
[Enhancement] Don't create not needed dirs in the non-Rails setup template.
[Enhancement] Improve printing of TTIN to separate threads.
[Enhancement] Support transactional ID resource in the Admin ACL management.
[Enhancement] Move post-forceful shutdown termination wait value to a config.
[Enhancement] Include consumer group, subscription group and other details in error logs for key error locations.
[Enhancement] Inherit from ActiveJob::QueueAdapters::AbstractAdapter when possible for ActiveJob base class.
[Enhancement] Disable Nagle algorithm by default for improved network performance.
[Maintenance] Add basic direct DD integration spec via DD gem karafka monitoring feature.
[Refactoring] Comprehensive Admin module refactoring: Extract topic operations into Admin::Topics class and consumer group operations into Admin::ConsumerGroups class with proper inheritance hierarchy, cross-class method usage optimization, and constants moved to appropriate locations where they are actually used.
[Refactoring] Move routing-related contracts from Karafka::Contracts:: to Karafka::Routing::Contracts:: namespace and reorganize error message structure in YAML files under routing: scope for better code organization and logical grouping.
[Refactoring] Move config-related contracts from Karafka::Contracts::Config to Karafka::Setup::Contracts::Config namespace and reorganize error message structure in YAML files under setup: scope for better code organization and logical grouping.
[Refactoring] Move CLI server contracts from Karafka::Contracts::ServerCliOptions to Karafka::Cli::Contracts::Server namespace and reorganize error message structure in YAML files under cli: scope for improved naming consistency and logical grouping.
[Fix] Improve same timestamp dispatch in scheduled messages on Ruby 3.2.
[Fix] Fix incorrect (6 seconds vs 60 seconds) reset of connections on non-recoverable errors.
[Fix] Introduce mutex-safe and thread-safe #inspect where needed.
[Fix] Fix too loose requirement of Ruby 3.0 when it was 3.1 via transitive dependencies.
[Fix] Fix Pro Cleaner Messages compatibility with external libraries that prepend modules to #each method (e.g., DataDog tracing).
[Change] Require karafka-rdkafka >= 0.21.0 to support new features.
[Change] Remove no longer needed cooperative.sticky rebalance patch.
[Change] Normalize how libs and deps are required (no functional change for the end user)
[Change] Remove Ruby 3.1 specs according to the EOL schedule.

[Breaking] Change how consistency of DLQ dispatches works in Pro (partition_key vs. direct partition id mapping).
[Breaking] Remove the headers source_key from the Pro DLQ dispatched messages as the original key is now fully preserved.
[Breaking] Use DLQ and Piping prefix source_ instead of original_ to align with naming convention of Kafka Streams and Apache Flink for future usage.
[Breaking] Rename scheduled jobs topics names in their config (Pro).
[Breaking] Change K8s listener response from 204 to 200 and include JSON body with reasons.
[Breaking] Replace admin config max_attempts with max_retries_duration and
[Feature] Parallel Segments for concurrent processing of the same partition with more than partition count of processes (Pro).
[Enhancement] Normalize topic + partition logs format.
[Enhancement] Support KIP-82 (header values of arrays).
[Enhancement] Enhance errors tracker with #counts that contains per-error class specific counters for granular flow handling.
[Enhancement] Provide explicit Karafka::Admin.copy_consumer_group API.
[Enhancement] Return explicit value from Karafka::Admin.copy_consumer_group and Karafka::Admin.rename_consumer_group APIs.
[Enhancement] Introduce balanced non-consistent VP distributor improving the utilization up to 50% (Pro).
[Enhancement] Make the error tracker for advanced DLQ strategies respond to #topic and #partition for context aware dispatches.
[Enhancement] Allow setting the workers thread priority and set it to -1 (50ms) by default.
[Enhancement] Enhance low-level client.pause event with timeout value (if provided).
[Enhancement] Introduce #marking_cursor API (defaults to #cursor) in the filtering API (Pro).
[Enhancement] Support multiple DLQ target topics via context aware strategies (Pro).
[Enhancement] Raise error when post-transactional committing of offset is done outside of the transaction (Pro).
[Enhancement] Include info level rebalance logger listener data.
[Enhancement] Include info level subscription start info.
[Enhancement] Make the generic error handling in the LoggerListener more descriptive by logging also the error class.
[Enhancement] Allow marking older offsets to support advanced rewind capabilities.
[Enhancement] Change optional #seek reset offset flag default to true as false is almost never used and seek by default should move the internal consumer offset position as well.
[Enhancement] Include Swarm node ID in the swarm process tags.
[Enhancement] Replace internal usage of MD5 with SHA256 for FIPS.
[Enhancement] Improve OSS vs. Pro specs execution isolation.
[Enhancement] Preload librdkafka code prior to forking in the Swarm mode to save memory.
[Enhancement] Extract errors tracker class reference into an internal errors_tracker_class config option (Pro).
[Enhancement] Support rdkafka native kafka polling customization for admin.
[Enhancement] Customize the multiplexing scale delay (Pro) per consumer group (Pro).
[Enhancement] Set topic.metadata.refresh.interval.ms for default producer in dev to 5s to align with consumer setup.
[Enhancement] Alias -2 and -1 with latest and earliest for seeking.
[Enhancement] Allow for usage of latest and earliest in the Karafka::Pro::Iterator.
[Enhancement] Failures during topics migrate (and other subcommands) don't show what topic failed, and why it's invalid.
[Enhancement] Apply changes to topics configuration in atomic independent requests when using Declarative Topics.
[Enhancement] Execute the help CLI command when no command provided (similar to Rails) to improve DX.
[Enhancement] Remove backtrace from the CLI error for incorrect commands (similar to Rails) to improve DX.
[Enhancement] Provide karafka topics help sub-help due to nesting of Declarative Topics actions.
[Enhancement] Use independent keys for different states of reporting in scheduled messages.
[Enhancement] Enrich scheduled messages state reporter with debug data.
[Enhancement] Introduce a new state called stopped to the scheduled messages.
[Enhancement] Do not overwrite the key in the Pro DLQ dispatched messages for routing reasons.
[Enhancement] Introduce errors_tracker.trace_id for distributed error details correlation with the Web UI.
[Enhancement] Improve contracts validations reporting.
[Enhancement] Optimize topic creation and repartitioning admin operations for topics with hundreds of partitions.
[Refactor] Introduce a bin/verify_kafka_warnings script to clean Kafka from temporary test-suite topics.
[Refactor] Introduce a bin/verify_topics_naming script to ensure proper test topics naming convention.
[Refactor] Make sure all temporary topics have a it- prefix in their name.
[Refactor] Improve CI specs parallelization.
[Maintenance] Lower the Karafka::Admin poll_timeout to 50 ms to improve responsiveness of admin operations.
[Maintenance] Require karafka-rdkafka >= 0.19.5 due to usage of #rd_kafka_global_init, KIP-82, new producer caching engine and improvements to the partition_key assignments.
[Maintenance] Add Deimos routing patch into integration suite not to break it in the future.
[Maintenance] Remove Rails 7.0 specs due to upcoming EOL.
[Fix] Fix Recurring Tasks and Scheduled Messages not working with Swarm (using closed producer).
[Fix] Fix a case where unknown_topic_or_part error could leak out of the consumer on consumer shutdown.
[Fix] Fix missing virtual_partitions.partitioner.error custom error logging in the LoggerListener.
[Fix] Prevent applied system filters #timeout from potentially interacting with user filters.
[Fix] Use more sane value in Admin#seek_consumer_group for long ago.
[Fix] Prevent multiplexing of 1:1 from routing.
[Fix] WaterDrop level aborting transaction may cause seek offset to move (Pro).
[Fix] Fix inconsistency in the logs where Karafka::Server originating logs would not have server id reference.
[Fix] Fix inconsistency in the logs where OS signal originating logs would not have server id reference.
[Fix] Post-fork WaterDrop instance looses some of the non-kafka settings.
[Fix] Max epoch tracking for early cleanup causes messages to be skipped until reload.
[Fix] optparse double parse loses ARGV.
[Fix] karafka cannot be required without Bundler.
[Fix] Scheduled Messages re-seek moves to latest on inheritance of initial offset when 0 offset is compacted.
[Fix] Seek to :latest without topic_partition_position (-1) will not seek at all.
[Fix] Extremely high turn over of scheduled messages can cause them not to reach EOF/Loaded state.
[Fix] Fix incorrectly passed max_wait_time to rdkafka (ms instead of seconds) causing too long wait.
[Fix] Remove aggresive requerying of the Kafka cluster on topic creation/removal/altering.
[Change] Move to trusted-publishers and remove signing since no longer needed.

[Breaking] Change how consistency of DLQ dispatches works in Pro (partition_key vs. direct partition id mapping).
[Breaking] Remove the headers source_key from the Pro DLQ dispatched messages as the original key is now fully preserved.
[Breaking] Use DLQ and Piping prefix source_ instead of original_ to align with naming convention of Kafka Streams and Apache Flink for future usage.
[Breaking] Rename scheduled jobs topics names in their config (Pro).
[Breaking] Change K8s listener response from 204 to 200 and include JSON body with reasons.
[Breaking] Replace admin config max_attempts with max_retries_duration and
[Feature] Parallel Segments for concurrent processing of the same partition with more than partition count of processes (Pro).
[Enhancement] Normalize topic + partition logs format.
[Enhancement] Support KIP-82 (header values of arrays).
[Enhancement] Enhance errors tracker with #counts that contains per-error class specific counters for granular flow handling.
[Enhancement] Provide explicit Karafka::Admin.copy_consumer_group API.
[Enhancement] Return explicit value from Karafka::Admin.copy_consumer_group and Karafka::Admin.rename_consumer_group APIs.
[Enhancement] Introduce balanced non-consistent VP distributor improving the utilization up to 50% (Pro).
[Enhancement] Make the error tracker for advanced DLQ strategies respond to #topic and #partition for context aware dispatches.
[Enhancement] Allow setting the workers thread priority and set it to -1 (50ms) by default.
[Enhancement] Enhance low-level client.pause event with timeout value (if provided).
[Enhancement] Introduce #marking_cursor API (defaults to #cursor) in the filtering API (Pro).
[Enhancement] Support multiple DLQ target topics via context aware strategies (Pro).
[Enhancement] Raise error when post-transactional committing of offset is done outside of the transaction (Pro).
[Enhancement] Include info level rebalance logger listener data.
[Enhancement] Include info level subscription start info.
[Enhancement] Make the generic error handling in the LoggerListener more descriptive by logging also the error class.
[Enhancement] Allow marking older offsets to support advanced rewind capabilities.
[Enhancement] Change optional #seek reset offset flag default to true as false is almost never used and seek by default should move the internal consumer offset position as well.
[Enhancement] Include Swarm node ID in the swarm process tags.
[Enhancement] Replace internal usage of MD5 with SHA256 for FIPS.
[Enhancement] Improve OSS vs. Pro specs execution isolation.
[Enhancement] Preload librdkafka code prior to forking in the Swarm mode to save memory.
[Enhancement] Extract errors tracker class reference into an internal errors_tracker_class config option (Pro).
[Enhancement] Support rdkafka native kafka polling customization for admin.
[Enhancement] Customize the multiplexing scale delay (Pro) per consumer group (Pro).
[Enhancement] Set topic.metadata.refresh.interval.ms for default producer in dev to 5s to align with consumer setup.
[Enhancement] Alias -2 and -1 with latest and earliest for seeking.
[Enhancement] Allow for usage of latest and earliest in the Karafka::Pro::Iterator.
[Enhancement] Failures during topics migrate (and other subcommands) don't show what topic failed, and why it's invalid.
[Enhancement] Apply changes to topics configuration in atomic independent requests when using Declarative Topics.
[Enhancement] Execute the help CLI command when no command provided (similar to Rails) to improve DX.
[Enhancement] Remove backtrace from the CLI error for incorrect commands (similar to Rails) to improve DX.
[Enhancement] Provide karafka topics help sub-help due to nesting of Declarative Topics actions.
[Enhancement] Use independent keys for different states of reporting in scheduled messages.
[Enhancement] Enrich scheduled messages state reporter with debug data.
[Enhancement] Introduce a new state called stopped to the scheduled messages.
[Enhancement] Do not overwrite the key in the Pro DLQ dispatched messages for routing reasons.
[Enhancement] Introduce errors_tracker.trace_id for distributed error details correlation with the Web UI.
[Enhancement] Improve contracts validations reporting.
[Enhancement] Optimize topic creation and repartitioning admin operations for topics with hundreds of partitions.
[Refactor] Introduce a bin/verify_kafka_warnings script to clean Kafka from temporary test-suite topics.
[Refactor] Introduce a bin/verify_topics_naming script to ensure proper test topics naming convention.
[Refactor] Make sure all temporary topics have a it- prefix in their name.
[Refactor] Improve CI specs parallelization.
[Maintenance] Lower the Karafka::Admin poll_timeout to 50 ms to improve responsiveness of admin operations.
[Maintenance] Require karafka-rdkafka >= 0.19.5 due to usage of #rd_kafka_global_init, KIP-82, new producer caching engine and improvements to the partition_key assignments.
[Maintenance] Add Deimos routing patch into integration suite not to break it in the future.
[Maintenance] Remove Rails 7.0 specs due to upcoming EOL.
[Fix] Fix Recurring Tasks and Scheduled Messages not working with Swarm (using closed producer).
[Fix] Fix a case where unknown_topic_or_part error could leak out of the consumer on consumer shutdown.
[Fix] Fix missing virtual_partitions.partitioner.error custom error logging in the LoggerListener.
[Fix] Prevent applied system filters #timeout from potentially interacting with user filters.
[Fix] Use more sane value in Admin#seek_consumer_group for long ago.
[Fix] Prevent multiplexing of 1:1 from routing.
[Fix] WaterDrop level aborting transaction may cause seek offset to move (Pro).
[Fix] Fix inconsistency in the logs where Karafka::Server originating logs would not have server id reference.
[Fix] Fix inconsistency in the logs where OS signal originating logs would not have server id reference.
[Fix] Post-fork WaterDrop instance looses some of the non-kafka settings.
[Fix] Max epoch tracking for early cleanup causes messages to be skipped until reload.
[Fix] optparse double parse loses ARGV.
[Fix] karafka cannot be required without Bundler.
[Fix] Scheduled Messages re-seek moves to latest on inheritance of initial offset when 0 offset is compacted.
[Fix] Seek to :latest without topic_partition_position (-1) will not seek at all.
[Fix] Extremely high turn over of scheduled messages can cause them not to reach EOF/Loaded state.
[Fix] Fix incorrectly passed max_wait_time to rdkafka (ms instead of seconds) causing too long wait.
[Fix] Remove aggresive requerying of the Kafka cluster on topic creation/removal/altering.
[Change] Move to trusted-publishers and remove signing since no longer needed.

[Breaking] Change how consistency of DLQ dispatches works in Pro (partition_key vs. direct partition id mapping).
[Breaking] Remove the headers source_key from the Pro DLQ dispatched messages as the original key is now fully preserved.
[Breaking] Use DLQ and Piping prefix source_ instead of original_ to align with naming convention of Kafka Streams and Apache Flink for future usage.
[Breaking] Rename scheduled jobs topics names in their config (Pro).
[Feature] Parallel Segments for concurrent processing of the same partition with more than partition count of processes (Pro).
[Enhancement] Support KIP-82 (header values of arrays).
[Enhancement] Enhance errors tracker with #counts that contains per-error class specific counters for granular flow handling.
[Enhancement] Provide explicit Karafka::Admin.copy_consumer_group API.
[Enhancement] Return explicit value from Karafka::Admin.copy_consumer_group and Karafka::Admin.rename_consumer_group APIs.
[Enhancement] Introduce balanced non-consistent VP distributor improving the utilization up to 50% (Pro).
[Enhancement] Make the error tracker for advanced DLQ strategies respond to #topic and #partition for context aware dispatches.
[Enhancement] Allow setting the workers thread priority and set it to -1 (50ms) by default.
[Enhancement] Enhance low-level client.pause event with timeout value (if provided).
[Enhancement] Introduce #marking_cursor API (defaults to #cursor) in the filtering API (Pro).
[Enhancement] Support multiple DLQ target topics via context aware strategies (Pro).
[Enhancement] Raise error when post-transactional committing of offset is done outside of the transaction (Pro).
[Enhancement] Include info level rebalance logger listener data.
[Enhancement] Include info level subscription start info.
[Enhancement] Make the generic error handling in the LoggerListener more descriptive by logging also the error class.
[Enhancement] Allow marking older offsets to support advanced rewind capabilities.
[Enhancement] Change optional #seek reset offset flag default to true as false is almost never used and seek by default should move the internal consumer offset position as well.
[Enhancement] Include Swarm node ID in the swarm process tags.
[Enhancement] Replace internal usage of MD5 with SHA256 for FIPS.
[Enhancement] Improve OSS vs. Pro specs execution isolation.
[Enhancement] Preload librdkafka code prior to forking in the Swarm mode to save memory.
[Enhancement] Extract errors tracker class reference into an internal errors_tracker_class config option (Pro).
[Enhancement] Support rdkafka native kafka polling customization for admin.
[Enhancement] Customize the multiplexing scale delay (Pro) per consumer group (Pro).
[Enhancement] Set topic.metadata.refresh.interval.ms for default producer in dev to 5s to align with consumer setup.
[Enhancement] Alias -2 and -1 with latest and earliest for seeking.
[Enhancement] Allow for usage of latest and earliest in the Karafka::Pro::Iterator.
[Enhancement] Failures during topics migrate (and other subcommands) don't show what topic failed, and why it's invalid.
[Enhancement] Apply changes to topics configuration in atomic independent requests when using Declarative Topics.
[Enhancement] Execute the help CLI command when no command provided (similar to Rails) to improve DX.
[Enhancement] Remove backtrace from the CLI error for incorrect commands (similar to Rails) to improve DX.
[Enhancement] Provide karafka topics help sub-help due to nesting of Declarative Topics actions.
[Enhancement] Use independent keys for different states of reporting in scheduled messages.
[Enhancement] Enrich scheduled messages state reporter with debug data.
[Enhancement] Introduce a new state called stopped to the scheduled messages.
[Enhancement] Do not overwrite the key in the Pro DLQ dispatched messages for routing reasons.
[Enhancement] Introduce errors_tracker.trace_id for distributed error details correlation with the Web UI.
[Refactor] Introduce a bin/verify_kafka_warnings script to clean Kafka from temporary test-suite topics.
[Refactor] Introduce a bin/verify_topics_naming script to ensure proper test topics naming convention.
[Refactor] Make sure all temporary topics have a it- prefix in their name.
[Refactor] Improve CI specs parallelization.
[Maintenance] Lower the Karafka::Admin poll_timeout to 50 ms to improve responsiveness of admin operations.
[Maintenance] Require karafka-rdkafka >= 0.19.5 due to usage of #rd_kafka_global_init, KIP-82, new producer caching engine and improvements to the partition_key assignments.
[Maintenance] Add Deimos routing patch into integration suite not to break it in the future.
[Maintenance] Remove Rails 7.0 specs due to upcoming EOL.
[Fix] Fix Recurring Tasks and Scheduled Messages not working with Swarm (using closed producer).
[Fix] Fix a case where unknown_topic_or_part error could leak out of the consumer on consumer shutdown.
[Fix] Fix missing virtual_partitions.partitioner.error custom error logging in the LoggerListener.
[Fix] Prevent applied system filters #timeout from potentially interacting with user filters.
[Fix] Use more sane value in Admin#seek_consumer_group for long ago.
[Fix] Prevent multiplexing of 1:1 from routing.
[Fix] WaterDrop level aborting transaction may cause seek offset to move (Pro).
[Fix] Fix inconsistency in the logs where Karafka::Server originating logs would not have server id reference.
[Fix] Fix inconsistency in the logs where OS signal originating logs would not have server id reference.
[Fix] Post-fork WaterDrop instance looses some of the non-kafka settings.
[Fix] Max epoch tracking for early cleanup causes messages to be skipped until reload.
[Fix] optparse double parse loses ARGV.
[Fix] karafka cannot be required without Bundler.
[Fix] Scheduled Messages re-seek moves to latest on inheritance of initial offset when 0 offset is compacted.
[Fix] Seek to :latest without topic_partition_position (-1) will not seek at all.
[Fix] Extremely high turn over of scheduled messages can cause them not to reach EOF/Loaded state.
[Change] Move to trusted-publishers and remove signing since no longer needed.

[Breaking] Use DLQ and Piping prefix source_ instead of original_ to align with naming convention of Kafka Streams and Apache Flink for future usage.
[Breaking] Rename scheduled jobs topics names in their config (Pro).
[Feature] Parallel Segments for concurrent processing of the same partition with more than partition count of processes (Pro).
[Enhancement] Support KIP-82 (header values of arrays).
[Enhancement] Enhance errors tracker with #counts that contains per-error class specific counters for granular flow handling.
[Enhancement] Provide explicit Karafka::Admin.copy_consumer_group API.
[Enhancement] Return explicit value from Karafka::Admin.copy_consumer_group and Karafka::Admin.rename_consumer_group APIs.
[Enhancement] Introduce balanced non-consistent VP distributor improving the utilization up to 50% (Pro).
[Enhancement] Make the error tracker for advanced DLQ strategies respond to #topic and #partition for context aware dispatches.
[Enhancement] Allow setting the workers thread priority and set it to -1 (50ms) by default.
[Enhancement] Enhance low-level client.pause event with timeout value (if provided).
[Enhancement] Introduce #marking_cursor API (defaults to #cursor) in the filtering API (Pro).
[Enhancement] Support multiple DLQ target topics via context aware strategies (Pro).
[Enhancement] Raise error when post-transactional committing of offset is done outside of the transaction (Pro).
[Enhancement] Include info level rebalance logger listener data.
[Enhancement] Include info level subscription start info.
[Enhancement] Make the generic error handling in the LoggerListener more descriptive by logging also the error class.
[Enhancement] Allow marking older offsets to support advanced rewind capabilities.
[Enhancement] Change optional #seek reset offset flag default to true as false is almost never used and seek by default should move the internal consumer offset position as well.
[Enhancement] Include Swarm node ID in the swarm process tags.
[Enhancement] Replace internal usage of MD5 with SHA256 for FIPS.
[Enhancement] Improve OSS vs. Pro specs execution isolation.
[Enhancement] Preload librdkafka code prior to forking in the Swarm mode to save memory.
[Enhancement] Extract errors tracker class reference into an internal errors_tracker_class config option (Pro).
[Enhancement] Support rdkafka native kafka polling customization for admin.
[Enhancement] Customize the multiplexing scale delay (Pro) per consumer group (Pro).
[Enhancement] Set topic.metadata.refresh.interval.ms for default producer in dev to 5s to align with consumer setup.
[Enhancement] Alias -2 and -1 with latest and earliest for seeking.
[Enhancement] Allow for usage of latest and earliest in the Karafka::Pro::Iterator.
[Enhancement] Failures during topics migrate (and other subcommands) don't show what topic failed, and why it's invalid.
[Enhancement] Apply changes to topics configuration in atomic independent requests when using Declarative Topics.
[Enhancement] Execute the help CLI command when no command provided (similar to Rails) to improve DX.
[Enhancement] Remove backtrace from the CLI error for incorrect commands (similar to Rails) to improve DX.
[Enhancement] Provide karafka topics help sub-help due to nesting of Declarative Topics actions.
[Refactor] Introduce a bin/verify_kafka_warnings script to clean Kafka from temporary test-suite topics.
[Refactor] Introduce a bin/verify_topics_naming script to ensure proper test topics naming convention.
[Refactor] Make sure all temporary topics have a it- prefix in their name.
[Refactor] Improve CI specs parallelization.
[Maintenance] Lower the Karafka::Admin poll_timeout to 50 ms to improve responsiveness of admin operations.
[Maintenance] Require karafka-rdkafka >= 0.19.2 due to usage of #rd_kafka_global_init, KIP-82 and the new producer caching engine.
[Maintenance] Add Deimos routing patch into integration suite not to break it in the future.
[Maintenance] Remove Rails 7.0 specs due to upcoming EOL.
[Fix] Fix Recurring Tasks and Scheduled Messages not working with Swarm (using closed producer).
[Fix] Fix a case where unknown_topic_or_part error could leak out of the consumer on consumer shutdown.
[Fix] Fix missing virtual_partitions.partitioner.error custom error logging in the LoggerListener.
[Fix] Prevent applied system filters #timeout from potentially interacting with user filters.
[Fix] Use more sane value in Admin#seek_consumer_group for long ago.
[Fix] Prevent multiplexing of 1:1 from routing.
[Fix] WaterDrop level aborting transaction may cause seek offset to move (Pro).
[Fix] Fix inconsistency in the logs where Karafka::Server originating logs would not have server id reference.
[Fix] Fix inconsistency in the logs where OS signal originating logs would not have server id reference.
[Fix] Post-fork WaterDrop instance looses some of the non-kafka settings.
[Fix] Max epoch tracking for early cleanup causes messages to be skipped until reload.
[Fix] optparse double parse loses ARGV.
[Fix] karafka cannot be required without Bundler.
[Fix] Scheduled Messages re-seek moves to latest on inheritance of initial offset when 0 offset is compacted.
[Fix] Seek to :latest without topic_partition_position (-1) will not seek at all.
[Change] Move to trusted-publishers and remove signing since no longer needed.

[Breaking] Use DLQ and Piping prefix source_ instead of original_ to align with naming convention of Kafka Streams and Apache Flink for future usage.
[Breaking] Rename scheduled jobs topics names in their config (Pro).
[Feature] Parallel Segments for concurrent processing of the same partition with more than partition count of processes (Pro).
[Enhancement] Support KIP-82 (header values of arrays).
[Enhancement] Enhance errors tracker with #counts that contains per-error class specific counters for granular flow handling.
[Enhancement] Provide explicit Karafka::Admin.copy_consumer_group API.
[Enhancement] Return explicit value from Karafka::Admin.copy_consumer_group and Karafka::Admin.rename_consumer_group APIs.
[Enhancement] Introduce balanced non-consistent VP distributor improving the utilization up to 50% (Pro).
[Enhancement] Make the error tracker for advanced DLQ strategies respond to #topic and #partition for context aware dispatches.
[Enhancement] Allow setting the workers thread priority and set it to -1 (50ms) by default.
[Enhancement] Enhance low-level client.pause event with timeout value (if provided).
[Enhancement] Introduce #marking_cursor API (defaults to #cursor) in the filtering API (Pro).
[Enhancement] Support multiple DLQ target topics via context aware strategies (Pro).
[Enhancement] Raise error when post-transactional committing of offset is done outside of the transaction (Pro).
[Enhancement] Include info level rebalance logger listener data.
[Enhancement] Include info level subscription start info.
[Enhancement] Make the generic error handling in the LoggerListener more descriptive by logging also the error class.
[Enhancement] Allow marking older offsets to support advanced rewind capabilities.
[Enhancement] Change optional #seek reset offset flag default to true as false is almost never used and seek by default should move the internal consumer offset position as well.
[Enhancement] Include Swarm node ID in the swarm process tags.
[Enhancement] Replace internal usage of MD5 with SHA256 for FIPS.
[Enhancement] Improve OSS vs. Pro specs execution isolation.
[Enhancement] Preload librdkafka code prior to forking in the Swarm mode to save memory.
[Enhancement] Extract errors tracker class reference into an internal errors_tracker_class config option (Pro).
[Enhancement] Support rdkafka native kafka polling customization for admin.
[Enhancement] Customize the multiplexing scale delay (Pro) per consumer group (Pro).
[Enhancement] Set topic.metadata.refresh.interval.ms for default producer in dev to 5s to align with consumer setup.
[Enhancement] Alias -2 and -1 with latest and earliest for seeking.
[Enhancement] Allow for usage of latest and earliest in the Karafka::Pro::Iterator.
[Refactor] Introduce a bin/verify_kafka_warnings script to clean Kafka from temporary test-suite topics.
[Refactor] Introduce a bin/verify_topics_naming script to ensure proper test topics naming convention.
[Refactor] Make sure all temporary topics have a it- prefix in their name.
[Refactor] Improve CI specs parallelization.
[Maintenance] Lower the Karafka::Admin poll_timeout to 50 ms to improve responsiveness of admin operations.
[Maintenance] Require karafka-rdkafka >= 0.19.2 due to usage of #rd_kafka_global_init, KIP-82 and the new producer caching engine.
[Maintenance] Add Deimos routing patch into integration suite not to break it in the future.
[Maintenance] Remove Rails 7.0 specs due to upcoming EOL.
[Fix] Fix Recurring Tasks and Scheduled Messages not working with Swarm (using closed producer).
[Fix] Fix a case where unknown_topic_or_part error could leak out of the consumer on consumer shutdown.
[Fix] Fix missing virtual_partitions.partitioner.error custom error logging in the LoggerListener.
[Fix] Prevent applied system filters #timeout from potentially interacting with user filters.
[Fix] Use more sane value in Admin#seek_consumer_group for long ago.
[Fix] Prevent multiplexing of 1:1 from routing.
[Fix] WaterDrop level aborting transaction may cause seek offset to move (Pro).
[Fix] Fix inconsistency in the logs where Karafka::Server originating logs would not have server id reference.
[Fix] Fix inconsistency in the logs where OS signal originating logs would not have server id reference.
[Fix] Post-fork WaterDrop instance looses some of the non-kafka settings.
[Fix] Max epoch tracking for early cleanup causes messages to be skipped until reload.
[Fix] optparse double parse loses ARGV.
[Fix] karafka cannot be required without Bundler.
[Fix] Scheduled Messages re-seek moves to latest on inheritance of initial offset when 0 offset is compacted.

[Fix] Make sure Bundler.with_unbundled_env is not called multiple times.

[Enhancement] Clean message key and headers when cleaning messages via the cleaner API (Pro).
[Enhancement] Allow for setting metadata: false in the cleaner API for granular cleaning control (Pro)
[Enhancement] Instrument successful transaction via consumer.consuming.transaction event (Pro).

Uh oh!

Releases: karafka/karafka

v2.5.2

Uh oh!

v2.5.1

Uh oh!

v2.5.1.beta1

Uh oh!

v2.5.0

Uh oh!

v2.5.0.rc2

Uh oh!

v2.5.0.rc1

Uh oh!

v2.5.0.beta2

Uh oh!

v2.5.0.beta1

Uh oh!

v2.4.18

Uh oh!

v2.4.17

Uh oh!