Kafka Troubleshooting¶
Overview¶
- Ops quick-reference skeleton; broker settings: Kafka Configuration
1. Metadata / connection failures¶
Symptoms¶
Connection to node -1 could not be established or unreachable brokers from Metadata.
Keywords¶
advertised.listeners / NAT / container networking
Steps¶
- Verify bootstrap addresses are reachable.
- Check
advertised.listenerson each broker. - Test every host:port returned in Metadata.
- Align security protocol with
listener.security.protocol.map.
Reference¶
2. Consumer lag keeps growing¶
Steps¶
- Check consumer process health and rebalance storms.
- Match partition count to consumer parallelism.
- Review
max.poll.interval.msand processing time. - Inspect broker disk/network and ISR shrink.
3. Under-Replicated Partitions (URP)¶
Steps¶
kafka-topics.sh --describefor URP partitions.- Check offline replica brokers (disk, GC, network).
- Tune
replica.lag.time.max.msfor cross-DC RTT. - Reassign partitions after replicas recover.
4. Disk full / retention issues¶
Steps¶
kafka-log-dirs.sh --describefor large topics.- Compare broker defaults vs topic-level retention overrides.
- Verify
cleanup.policyfor compact topics. - Emergency: expand disk or adjust retention per change process.