Your Kafka cluster shows UnderReplicatedPartitions > 0. Walk through your debugging steps.

Step 1: kafka-topics.sh --describe to identify which topics/partitions are under-replicated and on which brokers. Step 2: Check if the lagging broker is alive and healthy. Step 3: Check broker logs for errors (disk I/O, GC pauses, OOM, network issues). Step 4: Check replica lag: kafka-log-dirs.sh --describe to see per-replica offsets. Step 5: Check disk space — a full disk stops replication immediately. Step 6: Check network bandwidth between brokers. Step 7: If broker is healthy but lagging, check replica.lag.time.max.ms vs actual network/disk speed. Step 8: If caused by GC: tune JVM, reduce heap, switch to ZGC. Resolution: fix the root cause — URP should self-resolve once the follower catches up.

Expected key concepts: kafka-topics.sh, under-replicated, lagging broker, broker logs, GC pauses, OOM, disk space, kafka-log-dirs.sh, replica.lag.time.max.ms, ZGC, self-resolve