[
https://issues.apache.org/jira/browse/SPARK-50421?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dongjoon Hyun updated SPARK-50421:
----------------------------------
Fix Version/s: 3.5.4
> Executor memory config is incorrect when multiple stage level resource
> profiles works
> -------------------------------------------------------------------------------------
>
> Key: SPARK-50421
> URL: https://issues.apache.org/jira/browse/SPARK-50421
> Project: Spark
> Issue Type: Bug
> Components: Scheduler, YARN
> Affects Versions: 3.1.3, 3.2.4, 3.3.4, 3.4.4, 3.5.3
> Reporter: Terry Wang
> Assignee: Terry Wang
> Priority: Major
> Labels: pull-request-available
> Fix For: 4.0.0, 3.5.4
>
>
> https://issues.apache.org/jira/browse/SPARK-27495 has imported a great
> feature to help adjust stage's level resource profile(abbv rp), we are
> observing some config is not as expected when multiple resource profile
> works.
> When one stage assigned rp2 different with default rp, the
> `UnifiedMemoryManager` will still get old conf as the default rp, which will
> cause maxOffHeapMemory is still the old value!
> {code:scala}
> protected[this] val maxOffHeapMemory = conf.get(MEMORY_OFFHEAP_SIZE)
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
