@isunjin Ok, I think I was just connecting a different type of problem with the word "corruption", but the case makes sense because the assumption so far was probably that there are only global failover and that the `InputSplitAssigner` is reset. I think in this context the implementation is ok, I still not the biggest fan of pulling the `InputSplit` concept into `Execution` but at the same time I see that currently all alternatives that came to my mind have their own set of problems. Guess it would be ok for me to merge the PR, but I would like to ask @tillrohrmann for a second opinion.
[ Full content available at: https://github.com/apache/flink/pull/6684 ] This message was relayed via gitbox.apache.org for devnull@infra.apache.org
