Github user jerryshao commented on the issue:
https://github.com/apache/spark/pull/14916
Agree with @tgravescs .
Actually this issue only exists when local `yarn#client` process is gone
and application is killed by yarn command. In this case the staging dir will
not be cleaned up by AM.
If it is due to `EXIT_EXCEPTION_USER_CLASS `, then I think reattempt will
finally clean up the dir. Also relying on signal is vulnerable, since you
cannot distinguish why the signal is received.
So maybe writing some scripts to clean up these dirs periodically would be
much easier compared to fix it in Spark.
---
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]
