Github user mccheah commented on a diff in the pull request:
https://github.com/apache/spark/pull/4066#discussion_r23108057
--- Diff: core/src/main/scala/org/apache/spark/SparkHadoopWriter.scala ---
@@ -105,10 +107,20 @@ class SparkHadoopWriter(@transient jobConf: JobConf)
def commit() {
val taCtxt = getTaskContext()
val cmtr = getOutputCommitter()
+ val dagSchedulerActor =
+ AkkaUtils.makeDriverRef("DAGScheduler", SparkEnv.get.conf,
SparkEnv.get.actorSystem)
+ val askTimeout = AkkaUtils.askTimeout(SparkEnv.get.conf)
if (cmtr.needsTaskCommit(taCtxt)) {
try {
- cmtr.commitTask(taCtxt)
- logInfo (taID + ": Committed")
+ val canCommit: Boolean = AkkaUtils.askWithReply(
+ AskPermissionToCommitOutput(jobID, splitID, attemptID),
dagSchedulerActor, askTimeout)
--- End diff --
You could send a message from the driver node that launched the speculated
task to the executor that was running the original copy of the task. It would
be tricky to avoid a second race in this solution however.
---
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]
