Hi,we were running into the same problem that was described here[0] and are wondering if anyone is still working on/interested in this PR[1]. It would be great if this could be merged since we are collecting and distributing data of unknown sizes at the time of configuration and this leads to unexpected failures.
Thanks! Felix [0] https://issues.apache.org/jira/browse/FLINK-2603 [1] https://github.com/apache/flink/pull/934
