[
https://issues.apache.org/jira/browse/SPARK-60112?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
L. C. Hsieh updated SPARK-60112:
--------------------------------
Affects Version/s: 3.5.8
4.2.0
4.1.3
4.0.4
4.3.0
4.4.0
(was: 5.0.0)
> Track memory consumers by identity in TaskMemoryManager
> -------------------------------------------------------
>
> Key: SPARK-60112
> URL: https://issues.apache.org/jira/browse/SPARK-60112
> Project: Spark
> Issue Type: Improvement
> Components: Spark Core
> Affects Versions: 3.5.8, 4.2.0, 4.3.0, 4.1.3, 4.0.4, 4.4.0
> Reporter: L. C. Hsieh
> Priority: Major
>
> TaskMemoryManager keeps spillable consumers in a value-equality
> HashSet<MemoryConsumer>. A consumer whose equals is value-based (e.g. a Scala
> case class) is deduplicated against an equal but distinct consumer: the
> second instance is never offered for spilling and its memory is missing from
> the per-consumer memory breakdown on OOM. SPARK-60071 fixed this for
> HybridRowQueue. Tracking consumers in an identity-based set (e.g.
> Collections.newSetFromMap(new IdentityHashMap<>())) would protect every
> consumer in one place.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
