Hi Kostas, I'm confused about the implementation of the temporary workaround. Would it be possible to get a little more detail?
> you could simply not stop the job after the whole input is processed How does one determine when the job has processed the whole input? > then wait until the output is committed Isn't it necessary to manually commit the in-progress files rather than waiting? The reason being that the source task could have successfully completed and checkpoints would no longer be propagated downstream. Therefore, any bulk-encoded format would not commit. Best, Hubert -- Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/
