I agree with Jun here, that it would make it easier to do lag checking.
However, for individual checks it's really not that much trouble to do the
second request. If you're doing a lot of lag checking (like every consumer
and every topic) where the scale would start to make a difference, I would
Grant,
In addition to FetchRequest, currently we have another way to get the high
watermark through OffsetRequest (
https://cwiki.apache.org/confluence/display/KAFKA/A+Guide+To+The+Kafka+Protocol#AGuideToTheKafkaProtocol-OffsetRequest).
OffsetRequest is a read-only request and is much lighter
Here is an initial proposal to add HighwaterMarkOffset to the
OffsetFetchResponse:
https://cwiki.apache.org/confluence/display/KAFKA/KIP-17+-+Add+HighwaterMarkOffset+to+OffsetFetchResponse
I can add a jira and more implementation details if the
initial proposal has interest.
Thanks,
Grant
--
