educative.io

A bit of a confusion regarding kafka's in-sync replicas and high water marks

Since kafka’s in-sync-replicas are only considered “in-sync” if they are up to date with the leader’s data. Does’t that contradict with concept of the high-water mark? Aren’t leader nodes and in-sync-replicas by definition contain the same data? and hence should contain the same high-water mark (highest offset shared across all ISRs)?