Set partition.assignment.strategy=org.apache.kafka.clients.consumer.CooperativeStickyAssignor in consumer properties
Ensure all consumers in the group use the same assignor; mixed assignors cause fallback to eager rebalancing
Set session.timeout.ms and heartbeat.interval.ms conservatively to avoid spurious rebalances during rolling restarts
Use kafka-consumer-groups.sh --describe to observe partition assignments and rebalance counts
Monitor the kafka.consumer:type=consumer-coordinator-metrics,name=rebalance-rate-and-time JMX metric to confirm reduced rebalance frequency
Known gotchas
Rolling a group from EagerAssignor to CooperativeStickyAssignor requires all members to upgrade simultaneously or use the transitional CooperativeAndEagerAssignor for one cycle first
If any consumer crashes mid-rebalance, the incremental protocol can stall until session.timeout.ms elapses
Cooperative rebalancing requires Kafka client version 2.4+ and broker compatibility is not required but client version matters
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp