Github user vanzin commented on the pull request:
https://github.com/apache/spark/pull/5478#issuecomment-93779804
> if we install Spark on each node, then we can set
spark.executorEnv.PYTHONPATH
That's a workaround, but not really in the spirit of how Spark-on-YARN is
expected to work. For example, you'd have to have all nodes have `$SPARK_HOME`
be the same for that to work.
If this must be distributed as a separate file, then Sandy's solution is
the way to go. The pyspark zip needs to be treated the same way the spark
assembly is: if no configuration, find it locally and upload it to nodes using
the distributed cache; add a config option so that users can store that file on
HDFS or even use `local:` URIs in case they want to manually distribute the
file.
---
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]
