Hi all, I'd like to start a discussion on KIP-1372, which proposes rejecting topic creation and topic-level config changes that would set min.insync.replicas above the topic's replication factor, instead of silently accepting a configuration that can never satisfy acks=all (KAFKA-19726, and KAFKA-4680 from 2017).
KIP: https://cwiki.apache.org/confluence/spaces/KAFKA/pages/449286526/KIP-1372+Reject+topic+configurations+where+min.insync.replicas+exceeds+the+replication+factor This revisits ground covered by KIP-579 (2020, never adopted); the KIP explains why KIP-966's effective-min-ISR clamp changes that analysis. Feedback welcome. Gavin
