Please prioritize a proper long-term fix for this issue. it is a big scalability issue for high-parallelism job (e.g. over 1,000).
FLINK-10122 KafkaConsumer should use partitionable state over union state if partition discovery is not active On Fri, Sep 28, 2018 at 7:20 AM Till Rohrmann <trohrm...@apache.org> wrote: > Dear community, > > almost 2 months have passed since the Flink 1.6 release and some good > features for the next major release are close to completion. Therefore, I > would like to discuss a possible feature freeze for Flink 1.7. > > Given that we've released Flink 1.6 at the beginning of August, I would > like to aim for the end of October beginning of November for the 1.7 > release date. In order to have enough time for release testing, I would > therefore propose the 22nd of October as the feature freeze deadline. > > What do you think about this date (realistic, too early, too late) and > which features/issues do you think should get priority? > > Since every release needs a release manager, I would also volunteer for it. > > Cheers, > Till >