All Versions
68
Latest Version
Avg Release Cycle
47 days
Latest Release
43 days ago

Changelog History
Page 5

  • v0.5.0.1 Changes

    • 🛠 Fixed inconsistency in responders non-required topic definition. Now only required: false available
    • #101 - Responders fail when multiple_usage true and required false
    • 🛠 fix error on startup from waterdrop #102
    • Waterdrop 0.3.2.1 with kafka.hosts instead of kafka_hosts
    • #105 - Karafka::Monitor#caller_label not working with inherited monitors
    • #99 - Standalone mode (without Sidekiq)
    • #97 - Buffer responders single topics before send (pre-validation)
    • 👍 Better control over consumer thanks to additional config options
    • 👷 #111 - Dynamic worker assignment based on the income params
    • Long shutdown time fix

  • v0.4.2 Changes

    • 🚅 #87 - Re-consume mode with crone for better Rails/Rack integration
    • 🚚 Moved Karafka server related stuff into separate Karafka::Server class
    • 📇 Renamed Karafka::Runner into Karafka::Fetcher
    • Gem bump
    • ➕ Added chroot option to Zookeeper options
    • 🚚 Moved BROKERS_PATH into config from constant
    • ➕ Added Karafka consume CLI action for a short running single consumption round
    • 🛠 Small fixes to close broken connections
    • ⚡️ Readme updates

  • v0.4.1 Changes

    • 🚅 Explicit throw(:abort) required to halt before_enqueue (like in Rails 5)
    • #61 - autodiscovery of Kafka brokers based on Zookeeper data
    • #63 - Graceful shutdown with current offset state during data processing
    • #65 - Example of NewRelic monitor is outdated
    • #71 - Setup should be executed after user code is loaded
    • Gem bump x3
    • Rubocop remarks
    • 🚚 worker_timeout config option has been removed. It now needs to be defined manually by the framework user because WorkerGlass::Timeout can be disabled and we cannot use Karafka settings on a class level to initialize user code stuff
    • 🚚 Moved setup logic under setup/Setup namespace
    • 👍 Better defaults handling
    • #75 - Kafka and Zookeeper options as a hash
    • #82 - Karafka autodiscovery fails upon caching of configs
    • 🔧 #81 - Switch config management to dry configurable
    • 🔖 Version fix
    • ⬇️ Dropped support for Ruby 2.1.*
    • 💎 Ruby bump to 2.3.1

  • v0.4.0 Changes

    • ➕ Added WaterDrop gem with default configuration
    • ♻️ Refactoring of config logic to simplify adding new dependencies that need to be configured based on #setup data
    • Gem bump
    • ⚡️ Readme updates
    • 📇 Renamed cluster to actor_cluster for method names
    • 👷 Replaced SidekiqGlass with generic WorkerGlass lib
    • Application bootstrap in app.rb no longer required
    • ⚡️ Karafka.boot needs to be executed after all the application files are loaded (template updated)
    • ♻️ Small loader refactor (no API changes)
    • 0️⃣ Ruby 2.3.0 support (default)
    • No more rake tasks
    • Karafka CLI instead of rake tasks
    • 👷 Worker cli command allows passing additional options directly to Sidekiq
    • 📇 Renamed concurrency to max_concurrency - it describes better what happens - Karafka will use this number of threads only when required
    • ➕ Added wait_timeout that allows us to tune how long should we wait on a single socket connection (single topic) for new messages before going to next one (this applies to each thread separately)
    • Rubocop remarks
    • ✂ Removed Sinatra and Puma dependencies
    • Karafka Cli internal reorganization
    • Karafka Cli routes task
    • 📜 #37 - warn log for failed parsing of a message
    • #43 - wrong constant name
    • #44 - Method name conflict
    • #48 - Cannot load such file -- celluloid/current
    • #46 - Loading application
    • #45 - Set up monitor in config
    • #47 - rake karafka:run uses app.rb only
    • ⚡️ #53 - README update with Sinatra/Rails integration description
    • #41 - New Routing engine
    • 🚚 #54 - Move Karafka::Workers::BaseWorker to Karafka::BaseWorker
    • 👷 #55 - ApplicationController and ApplicationWorker

  • v0.3.2 Changes

    • 🔀 Karafka::Params::Params lazy load merge keys with string/symbol names priorities fix

  • v0.3.1 Changes

    • 📇 Renamed Karafka::Monitor to Karafka::Process to represent a Karafka process wrapper
    • ➕ Added Karafka::Monitoring that allows to add custom logging and monitoring with external libraries and systems
    • 🚚 Moved logging functionality into Karafka::Monitoring default monitoring
    • ➕ Added possibility to provide own monitoring as long as in responds to #notice and #notice_error
    • 🔊 Standardized logging format for all logs

  • v0.3.0 Changes

    • 👀 Switched from custom ParserError for each parser to general catching of Karafka::Errors::ParseError and its descendants
    • Gem bump
    • 🛠 Fixed #32 - now when using custom workers that does not inherit from Karafka::BaseWorker perform method is not required. Using custom workers means that the logic that would normally lie under #perform, needs to be executed directly from the worker.
    • 🛠 Fixed #31 - Technically didn't fix because this is how Sidekiq is meant to work, but provided possibility to assign custom interchangers that allow to bypass JSON encoding issues by converting data that goes to Redis to a required format (and parsing it back when it is fetched)
    • Added full parameters lazy load - content is no longer loaded during #perform_async if params are not used in before_enqueue
    • 0️⃣ No more namespaces for Redis by default (use separate DBs)

  • v0.1.21 Changes

    • Sidekiq 4.0.1 bump
    • Gem bump
    • ➕ Added direct celluloid requirement to Karafka (removed from Sidekiq)

  • v0.1.19 Changes

    • ⏱ Internal call - schedule naming change
    • Enqueue to perform_async naming in controller to follow Sidekiqs naming convention
    • Gem bump

  • v0.1.18 Changes

    • 🔄 Changed Redis configuration options into a single hash that is directly passed to Redis setup for Sidekiq
    • ➕ Added config.ru to provide a Sidekiq web UI (see README for more details)