CI tpcds q14b flaky failure on spark 3.0/3.1/3.2

[yew1eb]

CI TPCDS test q14b frequently fails intermittently (flaky failure) on Spark 3.0/3.1/3.2 versions, while it runs stably on higher Spark versions. ( Example failure job: link)

Reproduction

The failure is reproducible locally on Spark 3.1.

Failure Details

Failure occurs at stage 158, related to the final subquery and join in the q14b SQL:

Failure is in stage 158, related to the final subquery and join in the SQL:

Suspected Root Cause

Compatibility issues of the q14b final share HAVING filter subquery and join with Spark 3.0/3.1.

[yew1eb]

@cxzl25 @richox TPCDS q14b is stably failing on Spark 3.0/3.1 in CI, blocking PR merges.
Shall we skip it temporarily while investigating the root cause?

[cxzl25]

I also saw on CI that Spark3.2 will also fail with probability.
It looks like the DirectTaskResult#accumUpdates returned has null values.

https://github.com/apache/auron/actions/runs/21172720953/job/60896070881#step:21:2744

ERROR TaskResultGetter: Exception while getting task result
java.lang.NullPointerException
	at org.apache.spark.scheduler.TaskResultGetter$$anon$3.$anonfun$run$3(TaskResultGetter.scala:109)
	at scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:238)
	at scala.collection.mutable.ResizableArray.foreach(ResizableArray.scala:62)
	at scala.collection.mutable.ResizableArray.foreach$(ResizableArray.scala:55)
	at scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:49)
	at scala.collection.TraversableLike.map(TraversableLike.scala:238)
	at scala.collection.TraversableLike.map$(TraversableLike.scala:231)
	at scala.collection.AbstractTraversable.map(Traversable.scala:108)
	at org.apache.spark.scheduler.TaskResultGetter$$anon$3.$anonfun$run$1(TaskResultGetter.scala:108)
	at scala.runtime.java8.JFunction0$mcV$sp.apply(JFunction0$mcV$sp.java:23)
	at org.apache.spark.util.Utils$.logUncaughtExceptions(Utils.scala:1996)
	at org.apache.spark.scheduler.TaskResultGetter$$anon$3.run(TaskResultGetter.scala:63)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
	at java.lang.Thread.run(Thread.java:750)

[cxzl25]

Shall we skip it temporarily while investigating the root cause

OK