adixitconfluent commented on PR #17870: URL: https://github.com/apache/kafka/pull/17870#issuecomment-3056691649
Hi @showuon > 1. public API change seems to have no discussion with the community, nor no clear documentation. For share fetch, the broker is in charge of which data is returned. So, having the consumer trying to dictate the data for individual partitions did not fit neatly with that. We treated it as a small change to the API, hence did not have a discussion. Apologies in case we overstepped. > 2. I think this is a good improvement for the utilization of the fetch request, so why only in share fetch not normal fetch? Unlike SFR, fetch requests can fetch from all the partitions. Whether the replica manager provides new data or not, that's a separate discussion. In case of SFR, you don't even make the call to provide data from replica manager for the topic partitions that cannot be acquired. Having said that, maybe we can think of a logic to perform a retry in both fetch and share fetch in case the fetchable partitions are not returning enough data from replica manager. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
