Github user zentol commented on the pull request:
https://github.com/apache/flink/pull/202#issuecomment-90921879
The timeout measures how long java or python are stuck in a blocking udp
operation. this generally means how long it takes for the python side to
compute one chunk of data. the java side sends a chunk, and then waits for the
next signal. if this takes too long, timeout. it's quite fickle to be honest,
and without a regular heartbeat one can always construct scenarios where it
will break the job.
i *think* tcp would cover it.
thanks for the hint to remove the shutdown hook, addressed.
---
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.
---
