Github user maasg commented on the pull request:
https://github.com/apache/spark/pull/4027#issuecomment-93029751
One of the issues this PR is solving is ensuring jobs could be forced to
spread over several nodes. This is particularly important for Spark Streaming
as parallelizing the work over several physical nodes improves throughput by
the increased available IOPS. Currently, Spark Streaming performance is like a
slot machine au reverse: on each deployment you get a different combination of
executors on nodes. If you're lucky, getting all different and you win, all the
same and you're in trouble. More on that behavior is documented here:
https://issues.apache.org/jira/browse/SPARK-4940
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
