Github user taroplus commented on the issue:
https://github.com/apache/spark/pull/17074
SPARK-18646 is the one I filed, SPARK-19675 is the one I filed. But it kept
getting closed again and again, so I linked my PR to SPARK-18646 which is still
open.
But I think I know what it means. The author of that jira suggested that
the fix would be to pass parent class loader to the constructor. If we call
ExecutorClassLoader.getParent via debugger or something, it returns
SystemClassLoader which is wrong and the person says that it should be the
'parent' class loader.
Original change (fa0524f), which introduced this issue, was not to pass
_parent_ to the constructor so that this class loader takes precedence over
parent class loader, however it looks to me that the change forgot to set
parent as null so if system class loader has something different from current
or parent class loader, problem like this happens.
I'm not sure if I explained well, .. let me know if you need more info
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
