Why this matters
- VaultCommerce rolling deploys triggered 30-second rebalance storms with eager protocol — fixed with cooperative sticky.
- During rebalance, partitions may be revoked before reassignment — handle revocation gracefully.
- Rebalance listener hooks let VaultCommerce flush in-flight work before partition loss.
- Frequent rebalances often indicate slow poll loops or unstable membership.
New consumer joins
Partition reassignment
Resume consumption
Eager vs cooperative
Eager (range, round-robin): all partitions revoked, then reassigned — full stop. Cooperative sticky: only moved partitions revoked incrementally. VaultCommerce sets partition.assignment.strategy=CooperativeStickyAssignor on all Java consumers.
Key points
- Rebalance — partition ownership transfer among group members
- CooperativeStickyAssignor — incremental rebalance (Kafka 2.4+)
- onPartitionsRevoked — callback to commit offsets before losing partitions
- group.instance.id — static member identity across restarts
- max.poll.interval.ms — max time between poll() calls before considered dead
The rest of this guide is part of Pro
Free covers every fresher guide in full. Pro unlocks Senior, Staff and Architect guides across every track.
Still to come · 2 sections
- Triggers and mitigation
- VaultCommerce rollout checklist
Test yourself
Answer these before moving on — recall is what makes it stick.
- How do you perform rebalancing and what are the pitfalls (stability, consumer downtime)?
- Describe a real-world example: designing an event-driven order processing pipeline with Kafka—what are the key design decisions?
- Explain the core concepts of Kafka Streams: KStream, KTable, state stores, and joins.