Github user vanzin commented on a diff in the pull request:
https://github.com/apache/spark/pull/4155#discussion_r23474074
--- Diff:
core/src/main/scala/org/apache/spark/scheduler/OutputCommitCoordinator.scala ---
@@ -0,0 +1,178 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.scheduler
+
+import scala.collection.mutable
+import scala.concurrent.duration.FiniteDuration
+
+import akka.actor.{PoisonPill, ActorRef, Actor}
+
+import org.apache.spark.{SparkConf, Logging}
+import org.apache.spark.util.{AkkaUtils, ActorLogReceive}
+
+private[spark] sealed trait OutputCommitCoordinationMessage extends
Serializable
+
+private[spark] case class StageStarted(stage: Int) extends
OutputCommitCoordinationMessage
+private[spark] case class StageEnded(stage: Int) extends
OutputCommitCoordinationMessage
+private[spark] case object StopCoordinator extends
OutputCommitCoordinationMessage
+
+private[spark] case class AskPermissionToCommitOutput(
+ stage: Int,
+ task: Long,
+ taskAttempt: Long)
+ extends OutputCommitCoordinationMessage
+
+private[spark] case class TaskCompleted(
+ stage: Int,
+ task: Long,
+ attempt: Long,
+ successful: Boolean)
+ extends OutputCommitCoordinationMessage
+
+/**
+ * Authority that decides whether tasks can commit output to HDFS.
+ *
+ * This lives on the driver, but the actor allows the tasks that commit
+ * to Hadoop to invoke it.
+ */
+private[spark] class OutputCommitCoordinator(conf: SparkConf) extends
Logging {
+
+ // Initialized by SparkEnv
+ var coordinatorActor: Option[ActorRef] = None
+ private val timeout = AkkaUtils.askTimeout(conf)
+ private val maxAttempts = AkkaUtils.numRetries(conf)
+ private val retryInterval = AkkaUtils.retryWaitMs(conf)
+
+ private type StageId = Int
+ private type TaskId = Long
+ private type TaskAttemptId = Long
+
+ private val authorizedCommittersByStage:
+ mutable.Map[StageId, mutable.Map[TaskId, TaskAttemptId]] =
mutable.HashMap()
+
+ def stageStart(stage: StageId) {
+ sendToActor(StageStarted(stage))
+ }
+ def stageEnd(stage: StageId) {
+ sendToActor(StageEnded(stage))
+ }
+
+ def canCommit(
+ stage: StageId,
+ task: TaskId,
+ attempt: TaskAttemptId): Boolean = {
+ askActor(AskPermissionToCommitOutput(stage, task, attempt))
+ }
+
+ def taskCompleted(
+ stage: StageId,
+ task: TaskId,
+ attempt: TaskAttemptId,
+ successful: Boolean) {
+ sendToActor(TaskCompleted(stage, task, attempt, successful))
+ }
+
+ def stop() {
--- End diff --
Minor, but I think it's slightly weird that this class mixes methods that
should only be called from the driver (such as `stop`) and methods that
executors can call safely. Perhaps a check here that this is only being called
on the driver side?
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
