Oct 26, 2026 · Backend Development

Kafka 4.2: Share Groups Unlock Per-Partition Parallelism, Streams Rebalancing Moves to Broker

Kafka 4.2 introduces Share Groups (KIP-932) allowing multiple consumers to process records from the same partition concurrently using fetch locks and explicit acknowledgments (ACCEPT, RELEASE, REJECT, RENEW), while Kafka Streams gains server-side rebalancing (KIP-1071) and dead-letter queues (KIP-1034); version 4.2.2 adds 33 fixes including exactly-once checkpoint cleanup, deadlock resolution, and latency metric corrections.

Dead Letter QueueExactly-OnceKIP-1034

0 likes · 17 min read

Kafka 4.2: Share Groups Unlock Per-Partition Parallelism, Streams Rebalancing Moves to Broker

LuTiao Programming

Sep 3, 2026 · Backend Development

Kafka 4.2 Share Groups: Can Kafka Replace RabbitMQ for Task Queues?

The author tests Kafka 4.2 Share Groups (Kafka Queues) by migrating a report-generation task queue from RabbitMQ, showing how Share Groups enable elastic consumer scaling beyond partition limits, manual acknowledgment modes (ACCEPT/RELEASE/REJECT/RENEW), and long-task handling with renew(), while noting RabbitMQ's routing strengths remain relevant.

KafkaKafka QueuesMessage Queue

0 likes · 16 min read