What is the High Watermark (HWM)?

The High Watermark is the offset of the last message that has been fully replicated to all in-sync replicas (ISR). Consumers can only read messages up to and including the HWM — they cannot read ahead of it. This prevents consumers from reading data that might be rolled back if the leader fails before replication completes. The Log End Offset (LEO) marks the next offset to be written — the gap between LEO and HWM represents uncommitted data.

Expected key concepts: High Watermark, offset, replicated, ISR, consumers, LEO, Log End Offset, uncommitted, rolled back