Hello Richard, Thanks for the proposed KIP. I have a couple of general comments:
1. I'm not sure if piggy-backing the timeout exception on the existing requestTimeoutMs configured in "request.timeout.ms" is a good idea since a) it is a general config that applies for all types of requests, and 2) using it to cover all the phases of an API call, including network round trip and potential metadata refresh is shown to not be a good idea, as illustrated in KIP-91: https://cwiki.apache.org/confluence/display/KAFKA/KIP-91+Provide+Intuitive+User+Timeouts+in+The+Producer In fact, I think in KAFKA-4879 which is aimed for the same issue as KAFKA-6608, Jason has suggested we use a new config for the API. Maybe this would be a more intuitive manner than reusing the request.timeout.ms config. 2. Besides the Consumer.position() call, there are a couple of more blocking calls today that could result in infinite blocking: Consumer.commitSync() and Consumer.committed(), should they be considered in this KIP as well? 3. There are a few other APIs that are today relying on request.timeout.ms already for breaking the infinite blocking, namely Consumer.partitionFor(), Consumer.OffsetAndTimestamp() and Consumer.listTopics(), if we are making the other blocking calls to be relying a new config as suggested in 1) above, should we also change the semantics of these API functions for consistency? Guozhang On Sun, Mar 4, 2018 at 11:13 AM, Richard Yu <yohan.richard...@gmail.com> wrote: > Hi all, > > I would like to discuss a potential change which would be made to > KafkaConsumer: > https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=75974886 > > Thanks, > Richard Yu > -- -- Guozhang