Github user liancheng commented on a diff in the pull request:
https://github.com/apache/spark/pull/8687#discussion_r39151283
--- Diff: core/src/main/scala/org/apache/spark/rdd/PairRDDFunctions.scala
---
@@ -1051,6 +1049,15 @@ class PairRDDFunctions[K, V](self: RDD[(K, V)])
val jobAttemptId = newTaskAttemptID(jobtrackerID, stageId, isMap =
true, 0, 0)
val jobTaskContext = newTaskAttemptContext(wrappedConf.value,
jobAttemptId)
val jobCommitter = jobFormat.getOutputCommitter(jobTaskContext)
+
+ // If speculation is enabled, we only allow FileOutputCommitter.
+ val speculationEnabled = self.conf.getBoolean("spark.speculation",
false)
+ if (speculationEnabled && jobCommitter.getClass !=
+ classOf[org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter])
{
+ throw new SparkException(s"Cannot use ${jobCommitter.getClass} as
output committer when" +
+ "speculation is enabled.")
+ }
+
--- End diff --
I'm a little bit confused here. `jobCommitter` was retrieved via
`jobFormat.getOutputCommitter()`, which should return the default output
committer bundled with the output format. Does this mean
`saveAsNewAPIHadoopDataset()` never respects user-defined output committers? If
that's true, we should just leave this method as is. Restricting output
committer class to be `FileOutputCommitter` is too strong a limitation.
cc @yhuai
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
