Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/717#issuecomment-105502771
A min cutoff of 600 sounds a bit drastic.
I recall your experiments used an insanely large number of threads. Using a
proper batch scheduling mode for batch jobs is probably the right way to ensure
that we need less Thread space memory on average.
Should we delay this fix until we changed the scheduling and re-run your
experiments?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
