Github user twalthr commented on the issue:
https://github.com/apache/flink/pull/5811
Thanks for looking into it @kl0u. I observed the same behavior during
debugging. I will remove the check for now and open a follow up issue. If there
is no better solution, we might need to close the writer for checkpoints on
local filesystems for preventing data loss in cases where the OS/machine goes
down.---
