Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/292#issuecomment-70415876
A good piece of work!. I like that there is a much better test coverage for
the Yarn cluster now.
@rmetzger Can you comment on a few open questions?
- There are two projects for yarn: "flink-yarn" and "flink-yarn-tests".
Can we combine them in one project? The reason why we put the tests in
"flink-scala" into a different project was only beause the Scala macros uses in
"flink-scala" caused problems otherwise. Since the Yarn project is not using
macros, I see no reason for the separation.
- The example for the YARN session should probably not refer to a local
file ("file:///...")
- There is a bit of a strange mechanism in computing the ports for the
application master (base port plus attempt-id or so). Can this be simplified?
Also, what happens if that port is already in use, per chance?
- How hard would it be to make the client parameter that starts a yarn
session for the job simpler? Something like `-y` rather than `-m yarn-cluser`?
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
