turboFei commented on a change in pull request #25795: [SPARK-29037][Core]
Spark gives duplicate result when an application was killed
URL: https://github.com/apache/spark/pull/25795#discussion_r324452508
##########
File path:
core/src/main/scala/org/apache/spark/internal/io/HadoopMapReduceCommitProtocol.scala
##########
@@ -164,7 +164,9 @@ class HadoopMapReduceCommitProtocol(
}
override def commitJob(jobContext: JobContext, taskCommits:
Seq[TaskCommitMessage]): Unit = {
- committer.commitJob(jobContext)
+ if (!dynamicPartitionOverwrite) {
+ committer.commitJob(jobContext)
+ }
Review comment:
It works well when we set
spark.hadoop.mapreduce.fileoutputcommitter.algorithm.version to 2.
But for the version 2, is there a probability that partial result was
produced when we kill an application, which is committing tasks?
----------------------------------------------------------------
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]
