Hi David, On Fri, May 2, 2025 at 11:07 AM David Arthur <david.art...@confluent.io.invalid> wrote: > DA5: Do we have a way to quantify the metadata propagation latency? In the > past, I think we have eye-balled the metadata LEO between brokers and the > active controller to approximate the lag, but maybe you've found a better > way?
Kafka has a few important metrics for this defined in KIP-835: Monitor KRaft Controller Quorum Health. I think the metric you are looking for is MetadataLastAppliedRecordLagMs and last-applied-record-lag-ms. > DA6: Will the MV be the only way to enable/disable this new behavior? KRaft doesn't use MV for its features. This is because the MV is stored in the cluster metadata partition managed by KRaft. KRaft instead negotiates the RPC versions using the response from API_VERSIONS. Thanks, -- -José
