attilapiros commented on a change in pull request #24178:
[SPARK-25196][SQL][FOLLOWUP] Add synchronized for
InMemoryRelation.statsOfPlanToCache
URL: https://github.com/apache/spark/pull/24178#discussion_r268317291
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/CacheManager.scala
##########
@@ -163,12 +161,15 @@ class CacheManager extends Logging {
val relation = cachedData.cachedRepresentation
val (rowCount, newColStats) =
CommandUtils.computeColumnStats(sparkSession, relation, column)
- val oldStats = relation.statsOfPlanToCache
- val newStats = oldStats.copy(
- rowCount = Some(rowCount),
- attributeStats = AttributeMap((oldStats.attributeStats ++
newColStats).toSeq)
- )
- relation.statsOfPlanToCache = newStats
+
+ relation.synchronized {
Review comment:
I would move this logic into the class `InMemoryRelation`. Like:
```scala
def updateStats(rowCount: Long, newColStats: Map[Attribute, ColumnStat]):
Unit = synchronized {
val newStats = statsOfPlanToCache.copy(
rowCount = Some(rowCount),
attributeStats = AttributeMap((statsOfPlanToCache.attributeStats ++
newColStats).toSeq)
)
statsOfPlanToCache = newStats
}
```
This way both the write and the `synchronized` is in the member function of
the `InMemoryRelation` (I think it is bit more readable). What is your opinion?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
