[SPARK-23207] Shuffle+Repartition on an DataFrame could lead to incorrect answers - ASF JIRA

XML

Word

Printable

JSON

Details

Type: Bug
Status: Resolved
Priority: Blocker
Resolution: Fixed
Affects Version/s: 1.6.0, 2.0.0, 2.1.0, 2.2.0, 2.3.0
Fix Version/s: 2.1.4, 2.2.3, 2.3.0
Component/s: SQL
Labels:
- correctness

Target Version/s:

2.3.0

Description

Currently shuffle repartition uses RoundRobinPartitioning, the generated result is nondeterministic since the sequence of input rows are not determined.

The bug can be triggered when there is a repartition call following a shuffle (which would lead to non-deterministic row ordering), as the pattern shows below:
upstream stage -> repartition stage -> result stage
(-> indicate a shuffle)
When one of the executors process goes down, some tasks on the repartition stage will be retried and generate inconsistent ordering, and some tasks of the result stage will be retried generating different data.

The following code returns 931532, instead of 1000000:

import scala.sys.process._

import org.apache.spark.TaskContext
val res = spark.range(0, 1000 * 1000, 1).repartition(200).map { x =>
  x
}.repartition(200).map { x =>
  if (TaskContext.get.attemptNumber == 0 && TaskContext.get.partitionId < 2) {
    throw new Exception("pkill -f java".!!)
  }
  x
}
res.distinct().count()