ตัวรับฟังการปรับสมดุลใหม่และสมาชิกแบบคงที่

A rebalance reassigns partitions among consumers in a group. It happens when a consumer joins, leaves, crashes, or its session times out.

During a rebalance, processing pauses across the group.

Frequent rebalances hurt throughput: partitions are revoked and reassigned, in-flight work may be reprocessed, and local caches are invalidated.

Advanced consumers actively minimize and react to rebalances.

A ConsumerRebalanceListener gives you two callbacks:

• onPartitionsRevoked — called before partitions are taken away.
• onPartitionsAssigned — called after new partitions are granted.

Use onPartitionsRevoked to commit offsets for work completed so far, avoiding duplicate processing after reassignment.

In Spring Kafka, implement ConsumerAwareRebalanceListener and set it on the container properties.

Static membership assigns each consumer a stable group.instance.id. On a brief restart, the broker keeps its partitions instead of triggering a rebalance.

Static membership relies on session.timeout.ms. If a consumer rejoins within the session window using the same instance id, no rebalance occurs.

The CooperativeStickyAssignor reassigns only the partitions that must move, instead of revoking everything (stop-the-world). This dramatically reduces rebalance impact.

If processing a batch exceeds max.poll.interval.ms, the broker assumes the consumer is dead and rebalances. Keep per-poll work bounded or raise the interval.

For resilient consumers, combine static membership (avoid restart churn), cooperative rebalancing (smaller moves), and a rebalance listener (clean offset commits).

Rebalances are unavoidable but controllable. Listeners let you commit cleanly, static membership avoids needless rebalances on restart, and cooperative assignment limits disruption.

Test your understanding of rebalancing.

You learned advanced rebalance handling.

• Rebalances pause processing and can reprocess work.
• Rebalance listeners let you commit offsets on revoke.
• Static membership avoids rebalances on brief restarts.
• Cooperative assignment reduces stop-the-world disruption.