Apache Kafka

Consumer Fetch Latency Spike from Broker Overload

warning
latencyUpdated Mar 2, 2026

When consumer fetch latency increases significantly, it indicates the broker is slow to respond to fetch requests, often due to disk I/O, CPU saturation, or competing produce traffic.

Technologies:
How to detect:

Monitor kafka.consumer.fetch_latency_avg or kafka.request.fetch_consumer_time_avg exceeding baseline by 2x. Cross-reference with kafka.request.handler_idle_percent and kafka.consumer.fetch_rate to identify if broker overload or consumer behavior change.

Recommended action:

1. Check broker load: Monitor CPU, disk I/O, and network on broker. 2. Review competing traffic: Check if produce traffic is interfering with fetch requests. 3. Increase broker capacity: Scale vertically or horizontally if overloaded. 4. Tune fetch.min.bytes: Adjust to allow batching more data per fetch. 5. Check disk performance: Slow disk reads increase fetch latency. 6. Monitor partition leadership: Unbalanced leadership can overload specific brokers.