New features:
- Add async context manager support for both Producer and Consumer. (pr #613 and #494 by @nimish)
- Upgrade to kafka-python version 2.0.0 and set it as non-strict
parameter. (issue #590 by @yumendy and #558 by @originalgremlin) - Make loop argument optional (issue #544)
- SCRAM-SHA-256 and SCRAM-SHA-512 support for SASL authentication (issue #571 and pr #588 by @SukiCZ)
- Added headers param to AIOKafkaProducer.send_and_wait (pr #553 by @megabotan)
- Add
consumer.last_poll_timestamp(partition)which gives the ms timestamp of the last
update ofhighwaterandlso. (issue #523 and pr #526 by @aure-olli) - Change all code base to async-await (pr #522)
- Minor: added PR and ISSUE templates to GitHub
Bugfixes:
- Ignore debug package generation on bdist_rpm command. (issue #599 by @gabriel-tincu)
- UnknownMemberId was raised to the user instead of retrying on auto commit. (issue #611)
- Fix issue with messages not being read after subscriptions change with group_id=None. (issue #536)
- Handle
RequestTimedOutErrorincoordinator._do_commit_offsets()method to explicitly mark
coordinator as dead. (issue #584 and pr #585 by @FedirAlifirenko) - Added handling
asyncio.TimeoutErroron metadata request to broker and metadata update.
(issue #576 and pr #577 by @MichalMazurek) - Too many reqs on kafka not available (issue #496 by @lud4ik)
- Consumer.seek_to_committed now returns mapping of committed offsets (pr #531 by @ask)
- Message Accumulator: add_message being recursive eventually overflows (pr #530 by @ask)
Improved Documentation:
- Clarify auto_offset_reset usage. (pr 601 by @dargor)
- Fix spelling errors in comments and documentation using codespell (pr #567 by mauritsvdvijgh)
- Delete old benchmark file (issue #546 by @jeffwidman)
- Fix a few typos in docs (pr #573 and pr #563 by @ultrabug)
- Fix typos, spelling, grammar, etc (pr #545 and pr #547 by @jeffwidman)
- Fix typo in docs (pr #541 by @pablogamboa)
- Fix documentation for benchmark (pr #537 by @abhishekray07)
- Better logging for bad CRC (pr #529 by @ask)