Kafunk v0.1.15 Release Notes

    • ๐Ÿ’ฅ BREAKING: FetchResponse item now explicit structure rather than tuple. Breaking only if using low-level Fetch API.

    • BUG: consumer doesn't recover from leaderless_partition_detected #204

    • BUG: consumer would enter an infinite loop between offsets_out_of_range and resuming_fetch_from_reset_offsets

    • ๐Ÿ› BUG: stalled consumers after TCP connection timeout

    • ๐Ÿ› BUG: offsets committed for partitions which are no longer assigned after a rebalance

    • ๐Ÿ”‹ FEATURE: ConsumerConfig.fetchMaxBytesOverride allows increasing fetchMaxBytes to a ceiling on message too big exceptions.

    • ๐Ÿ”‹ FEATURE: ProducerQueueType to allow configuration of producer message queue networks.

    • ๐Ÿ‘Œ IMPROVE: support for v0.11 of broker protocol, including v5 of Fetch and v3 of Produce APIs

    • IMPROVE: logging where received_empty_buffer logged as INFO message as it doesn't indicate an error

    • ๐Ÿ‘Œ IMPROVE: connection managemeng to ensure only a single TCP connection per broker per Kafka connection

    • ๐Ÿ‘Œ IMPROVE: Reduce recovery time from broker restarts and rebalancing

    • ๐Ÿ‘Œ IMPROVE: improved TCP connection logging and log levels to reduce "false positives"

    • ๐Ÿ‘Œ IMPROVE: .NET Standard support