[SPARK-1397] Notify SparkListeners when stages fail or are cancelled.

[I wanted to post this for folks to comment but it depends on (and thus includes the changes in) a currently outstanding PR, apache#305. You can look at just the second commit: kayousterhout@93f08ba to see just the changes relevant to this PR] Previously, when stages fail or get cancelled, the SparkListener is only notified indirectly through the SparkListenerJobEnd, where we sometimes pass in a single stage that failed. This worked before job cancellation, because jobs would only fail due to a single stage failure. However, with job cancellation, multiple running stages can fail when a job gets cancelled. Right now, this is not handled correctly, which results in stages that get stuck in the “Running Stages” window in the UI even though they’re dead. This PR changes the SparkListenerStageCompleted event to a SparkListenerStageEnded event, and uses this event to tell SparkListeners when stages fail in addition to when they complete successfully. This change is NOT publicly backward compatible for two reasons. First, it changes the SparkListener interface. We could alternately add a new event, SparkListenerStageFailed, and keep the existing SparkListenerStageCompleted. However, this is less consistent with the listener events for tasks / jobs ending, and will result in some code duplication for listeners (because failed and completed stages are handled in similar ways). Note that I haven’t finished updating the JSON code to correctly handle the new event because I’m waiting for feedback on whether this is a good or bad idea (hence the “WIP”). It is also not backwards compatible because it changes the publicly visible JobWaiter.jobFailed() method to no longer include a stage that caused the failure. I think this change should definitely stay, because with cancellation (as described above), a failure isn’t necessarily caused by a single stage. Author: Kay Ousterhout <[email protected]> Closes apache#309 from kayousterhout/stage_cancellation and squashes the following commits: 5533ecd [Kay Ousterhout] Fixes in response to Mark's review 320c7c7 [Kay Ousterhout] Notify SparkListeners when stages fail or are cancelled.
knowledgehacker · Apr 8, 2014 · fac6085 · fac6085
1 parent e25b593
commit fac6085
Show file tree

Hide file tree

Showing 11 changed files with 151 additions and 78 deletions.
diff --git a/core/src/main/scala/org/apache/spark/FutureAction.scala b/core/src/main/scala/org/apache/spark/FutureAction.scala
@@ -141,7 +141,7 @@ class SimpleFutureAction[T] private[spark](jobWaiter: JobWaiter[_], resultFunc:
   private def awaitResult(): Try[T] = {
     jobWaiter.awaitResult() match {
       case JobSucceeded => scala.util.Success(resultFunc)
-      case JobFailed(e: Exception, _) => scala.util.Failure(e)
+      case JobFailed(e: Exception) => scala.util.Failure(e)
     }
   }
 }

diff --git a/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala b/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
@@ -342,22 +342,24 @@ class DAGScheduler(
   }
 
   /**
-   * Removes job and any stages that are not needed by any other job.  Returns the set of ids for
-   * stages that were removed.  The associated tasks for those stages need to be cancelled if we
-   * got here via job cancellation.
+   * Removes state for job and any stages that are not needed by any other job.  Does not
+   * handle cancelling tasks or notifying the SparkListener about finished jobs/stages/tasks.
+   *
+   * @param job The job whose state to cleanup.
+   * @param resultStage Specifies the result stage for the job; if set to None, this method
+   *                    searches resultStagesToJob to find and cleanup the appropriate result stage.
    */
-  private def removeJobAndIndependentStages(jobId: Int): Set[Int] = {
-    val registeredStages = jobIdToStageIds(jobId)
-    val independentStages = new HashSet[Int]()
-    if (registeredStages.isEmpty) {
-      logError("No stages registered for job " + jobId)
+  private def cleanupStateForJobAndIndependentStages(job: ActiveJob, resultStage: Option[Stage]) {
+    val registeredStages = jobIdToStageIds.get(job.jobId)
+    if (registeredStages.isEmpty || registeredStages.get.isEmpty) {
+      logError("No stages registered for job " + job.jobId)
     } else {
-      stageIdToJobIds.filterKeys(stageId => registeredStages.contains(stageId)).foreach {
+      stageIdToJobIds.filterKeys(stageId => registeredStages.get.contains(stageId)).foreach {
         case (stageId, jobSet) =>
-          if (!jobSet.contains(jobId)) {
+          if (!jobSet.contains(job.jobId)) {
             logError(
               "Job %d not registered for stage %d even though that stage was registered for the job"
-              .format(jobId, stageId))
+              .format(job.jobId, stageId))
           } else {
             def removeStage(stageId: Int) {
               // data structures based on Stage
@@ -394,23 +396,28 @@ class DAGScheduler(
                 .format(stageId, stageIdToStage.size))
             }
 
-            jobSet -= jobId
+            jobSet -= job.jobId
             if (jobSet.isEmpty) { // no other job needs this stage
-              independentStages += stageId
               removeStage(stageId)
             }
           }
       }
     }
-    independentStages.toSet
-  }
+    jobIdToStageIds -= job.jobId
+    jobIdToActiveJob -= job.jobId
+    activeJobs -= job
 
-  private def jobIdToStageIdsRemove(jobId: Int) {
-    if (!jobIdToStageIds.contains(jobId)) {
-      logDebug("Trying to remove unregistered job " + jobId)
+    if (resultStage.isEmpty) {
+      // Clean up result stages.
+      val resultStagesForJob = resultStageToJob.keySet.filter(
+        stage => resultStageToJob(stage).jobId == job.jobId)
+      if (resultStagesForJob.size != 1) {
+        logWarning(
+          s"${resultStagesForJob.size} result stages for job ${job.jobId} (expect exactly 1)")
+      }
+      resultStageToJob --= resultStagesForJob
     } else {
-      removeJobAndIndependentStages(jobId)
-      jobIdToStageIds -= jobId
+      resultStageToJob -= resultStage.get
     }
   }
 
@@ -460,7 +467,7 @@ class DAGScheduler(
     val waiter = submitJob(rdd, func, partitions, callSite, allowLocal, resultHandler, properties)
     waiter.awaitResult() match {
       case JobSucceeded => {}
-      case JobFailed(exception: Exception, _) =>
+      case JobFailed(exception: Exception) =>
         logInfo("Failed to run " + callSite)
         throw exception
     }
@@ -606,7 +613,16 @@ class DAGScheduler(
         for (job <- activeJobs) {
           val error = new SparkException("Job cancelled because SparkContext was shut down")
           job.listener.jobFailed(error)
-          listenerBus.post(SparkListenerJobEnd(job.jobId, JobFailed(error, -1)))
+          // Tell the listeners that all of the running stages have ended.  Don't bother
+          // cancelling the stages because if the DAG scheduler is stopped, the entire application
+          // is in the process of getting stopped.
+          val stageFailedMessage = "Stage cancelled because SparkContext was shut down"
+          runningStages.foreach { stage =>
+            val info = stageToInfos(stage)
+            info.stageFailed(stageFailedMessage)
+            listenerBus.post(SparkListenerStageCompleted(info))
+          }
+          listenerBus.post(SparkListenerJobEnd(job.jobId, JobFailed(error)))
         }
         return true
     }
@@ -676,7 +692,7 @@ class DAGScheduler(
       }
     } catch {
       case e: Exception =>
-        jobResult = JobFailed(e, job.finalStage.id)
+        jobResult = JobFailed(e)
         job.listener.jobFailed(e)
     } finally {
       val s = job.finalStage
@@ -826,11 +842,8 @@ class DAGScheduler(
                   job.numFinished += 1
                   // If the whole job has finished, remove it
                   if (job.numFinished == job.numPartitions) {
-                    jobIdToActiveJob -= stage.jobId
-                    activeJobs -= job
-                    resultStageToJob -= stage
                     markStageAsFinished(stage)
-                    jobIdToStageIdsRemove(job.jobId)
+                    cleanupStateForJobAndIndependentStages(job, Some(stage))
                     listenerBus.post(SparkListenerJobEnd(job.jobId, JobSucceeded))
                   }
                   job.listener.taskSucceeded(rt.outputId, event.result)
@@ -982,7 +995,7 @@ class DAGScheduler(
     if (!jobIdToStageIds.contains(jobId)) {
       logDebug("Trying to cancel unregistered job " + jobId)
     } else {
-      failJobAndIndependentStages(jobIdToActiveJob(jobId), s"Job $jobId cancelled")
+      failJobAndIndependentStages(jobIdToActiveJob(jobId), s"Job $jobId cancelled", None)
     }
   }
 
@@ -999,7 +1012,8 @@ class DAGScheduler(
     stageToInfos(failedStage).completionTime = Some(System.currentTimeMillis())
     for (resultStage <- dependentStages) {
       val job = resultStageToJob(resultStage)
-      failJobAndIndependentStages(job, s"Job aborted due to stage failure: $reason")
+      failJobAndIndependentStages(job, s"Job aborted due to stage failure: $reason",
+        Some(resultStage))
     }
     if (dependentStages.isEmpty) {
       logInfo("Ignoring failure of " + failedStage + " because all jobs depending on it are done")
@@ -1008,28 +1022,45 @@ class DAGScheduler(
 
   /**
    * Fails a job and all stages that are only used by that job, and cleans up relevant state.
+   *
+   * @param resultStage The result stage for the job, if known. Used to cleanup state for the job
+   *                    slightly more efficiently than when not specified.
    */
-  private def failJobAndIndependentStages(job: ActiveJob, failureReason: String) {
+  private def failJobAndIndependentStages(job: ActiveJob, failureReason: String,
+      resultStage: Option[Stage]) {
     val error = new SparkException(failureReason)
     job.listener.jobFailed(error)
 
-    // Cancel all tasks in independent stages.
-    val independentStages = removeJobAndIndependentStages(job.jobId)
-    independentStages.foreach(taskScheduler.cancelTasks)
-
-    // Clean up remaining state we store for the job.
-    jobIdToActiveJob -= job.jobId
-    activeJobs -= job
-    jobIdToStageIds -= job.jobId
-    val resultStagesForJob = resultStageToJob.keySet.filter(
-      stage => resultStageToJob(stage).jobId == job.jobId)
-    if (resultStagesForJob.size != 1) {
-      logWarning(
-        s"${resultStagesForJob.size} result stages for job ${job.jobId} (expect exactly 1)")
+    // Cancel all independent, running stages.
+    val stages = jobIdToStageIds(job.jobId)
+    if (stages.isEmpty) {
+      logError("No stages registered for job " + job.jobId)
     }
-    resultStageToJob --= resultStagesForJob
+    stages.foreach { stageId =>
+      val jobsForStage = stageIdToJobIds.get(stageId)
+      if (jobsForStage.isEmpty || !jobsForStage.get.contains(job.jobId)) {
+        logError(
+          "Job %d not registered for stage %d even though that stage was registered for the job"
+            .format(job.jobId, stageId))
+      } else if (jobsForStage.get.size == 1) {
+        if (!stageIdToStage.contains(stageId)) {
+          logError("Missing Stage for stage with id $stageId")
+        } else {
+          // This is the only job that uses this stage, so fail the stage if it is running.
+          val stage = stageIdToStage(stageId)
+          if (runningStages.contains(stage)) {
+            taskScheduler.cancelTasks(stageId)
+            val stageInfo = stageToInfos(stage)
+            stageInfo.stageFailed(failureReason)
+            listenerBus.post(SparkListenerStageCompleted(stageToInfos(stage)))
+          }
+        }
+      }
+    }
+
+    cleanupStateForJobAndIndependentStages(job, resultStage)
 
-    listenerBus.post(SparkListenerJobEnd(job.jobId, JobFailed(error, job.finalStage.id)))
+    listenerBus.post(SparkListenerJobEnd(job.jobId, JobFailed(error)))
   }
 
   /**

diff --git a/core/src/main/scala/org/apache/spark/scheduler/JobLogger.scala b/core/src/main/scala/org/apache/spark/scheduler/JobLogger.scala
@@ -191,7 +191,11 @@ class JobLogger(val user: String, val logDirName: String) extends SparkListener
    */
   override def onStageCompleted(stageCompleted: SparkListenerStageCompleted) {
     val stageId = stageCompleted.stageInfo.stageId
-    stageLogInfo(stageId, "STAGE_ID=%d STATUS=COMPLETED".format(stageId))
+    if (stageCompleted.stageInfo.failureReason.isEmpty) {
+      stageLogInfo(stageId, s"STAGE_ID=$stageId STATUS=COMPLETED")
+    } else {
+      stageLogInfo(stageId, s"STAGE_ID=$stageId STATUS=FAILED")
+    }
   }
 
   /**
@@ -227,7 +231,7 @@ class JobLogger(val user: String, val logDirName: String) extends SparkListener
     var info = "JOB_ID=" + jobId
     jobEnd.jobResult match {
       case JobSucceeded => info += " STATUS=SUCCESS"
-      case JobFailed(exception, _) =>
+      case JobFailed(exception) =>
         info += " STATUS=FAILED REASON="
         exception.getMessage.split("\\s+").foreach(info += _ + "_")
       case _ =>

diff --git a/core/src/main/scala/org/apache/spark/scheduler/JobResult.scala b/core/src/main/scala/org/apache/spark/scheduler/JobResult.scala
@@ -24,5 +24,4 @@ private[spark] sealed trait JobResult
 
 private[spark] case object JobSucceeded extends JobResult
 
-// A failed stage ID of -1 means there is not a particular stage that caused the failure
-private[spark] case class JobFailed(exception: Exception, failedStageId: Int) extends JobResult
+private[spark] case class JobFailed(exception: Exception) extends JobResult
diff --git a/core/src/main/scala/org/apache/spark/scheduler/JobWaiter.scala b/core/src/main/scala/org/apache/spark/scheduler/JobWaiter.scala
@@ -64,7 +64,7 @@ private[spark] class JobWaiter[T](
 
   override def jobFailed(exception: Exception): Unit = synchronized {
     _jobFinished = true
-    jobResult = JobFailed(exception, -1)
+    jobResult = JobFailed(exception)
     this.notifyAll()
   }
 

diff --git a/core/src/main/scala/org/apache/spark/scheduler/SparkListener.scala b/core/src/main/scala/org/apache/spark/scheduler/SparkListener.scala
@@ -71,7 +71,7 @@ private[spark] case object SparkListenerShutdown extends SparkListenerEvent
  */
 trait SparkListener {
   /**
-   * Called when a stage is completed, with information on the completed stage
+   * Called when a stage completes successfully or fails, with information on the completed stage.
    */
   def onStageCompleted(stageCompleted: SparkListenerStageCompleted) { }
 

diff --git a/core/src/main/scala/org/apache/spark/scheduler/StageInfo.scala b/core/src/main/scala/org/apache/spark/scheduler/StageInfo.scala
@@ -26,8 +26,17 @@ private[spark]
 class StageInfo(val stageId: Int, val name: String, val numTasks: Int, val rddInfo: RDDInfo) {
   /** When this stage was submitted from the DAGScheduler to a TaskScheduler. */
   var submissionTime: Option[Long] = None
+  /** Time when all tasks in the stage completed or when the stage was cancelled. */
   var completionTime: Option[Long] = None
+  /** If the stage failed, the reason why. */
+  var failureReason: Option[String] = None
+
   var emittedTaskSizeWarning = false
+
+  def stageFailed(reason: String) {
+    failureReason = Some(reason)
+    completionTime = Some(System.currentTimeMillis)
+  }
 }
 
 private[spark]

diff --git a/core/src/main/scala/org/apache/spark/ui/jobs/JobProgressListener.scala b/core/src/main/scala/org/apache/spark/ui/jobs/JobProgressListener.scala
@@ -74,8 +74,13 @@ private[ui] class JobProgressListener(conf: SparkConf) extends SparkListener {
     // Remove by stageId, rather than by StageInfo, in case the StageInfo is from storage
     poolToActiveStages(stageIdToPool(stageId)).remove(stageId)
     activeStages.remove(stageId)
-    completedStages += stage
-    trimIfNecessary(completedStages)
+    if (stage.failureReason.isEmpty) {
+      completedStages += stage
+      trimIfNecessary(completedStages)
+    } else {
+      failedStages += stage
+      trimIfNecessary(failedStages)
+    }
   }
 
   /** If stages is too large, remove and garbage collect old stages */
@@ -215,20 +220,6 @@ private[ui] class JobProgressListener(conf: SparkConf) extends SparkListener {
     }
   }
 
-  override def onJobEnd(jobEnd: SparkListenerJobEnd) = synchronized {
-    jobEnd.jobResult match {
-      case JobFailed(_, stageId) =>
-        activeStages.get(stageId).foreach { s =>
-          // Remove by stageId, rather than by StageInfo, in case the StageInfo is from storage
-          activeStages.remove(s.stageId)
-          poolToActiveStages(stageIdToPool(stageId)).remove(s.stageId)
-          failedStages += s
-          trimIfNecessary(failedStages)
-        }
-      case _ =>
-    }
-  }
-
   override def onEnvironmentUpdate(environmentUpdate: SparkListenerEnvironmentUpdate) {
     synchronized {
       val schedulingModeName =

diff --git a/core/src/main/scala/org/apache/spark/util/JsonProtocol.scala b/core/src/main/scala/org/apache/spark/util/JsonProtocol.scala
@@ -166,12 +166,14 @@ private[spark] object JsonProtocol {
     val rddInfo = rddInfoToJson(stageInfo.rddInfo)
     val submissionTime = stageInfo.submissionTime.map(JInt(_)).getOrElse(JNothing)
     val completionTime = stageInfo.completionTime.map(JInt(_)).getOrElse(JNothing)
+    val failureReason = stageInfo.failureReason.map(JString(_)).getOrElse(JNothing)
     ("Stage ID" -> stageInfo.stageId) ~
     ("Stage Name" -> stageInfo.name) ~
     ("Number of Tasks" -> stageInfo.numTasks) ~
     ("RDD Info" -> rddInfo) ~
     ("Submission Time" -> submissionTime) ~
     ("Completion Time" -> completionTime) ~
+    ("Failure Reason" -> failureReason) ~
     ("Emitted Task Size Warning" -> stageInfo.emittedTaskSizeWarning)
   }
 
@@ -259,9 +261,7 @@ private[spark] object JsonProtocol {
     val json = jobResult match {
       case JobSucceeded => Utils.emptyJson
       case jobFailed: JobFailed =>
-        val exception = exceptionToJson(jobFailed.exception)
-        ("Exception" -> exception) ~
-        ("Failed Stage ID" -> jobFailed.failedStageId)
+        JObject("Exception" -> exceptionToJson(jobFailed.exception))
     }
     ("Result" -> result) ~ json
   }
@@ -442,11 +442,13 @@ private[spark] object JsonProtocol {
     val rddInfo = rddInfoFromJson(json \ "RDD Info")
     val submissionTime = Utils.jsonOption(json \ "Submission Time").map(_.extract[Long])
     val completionTime = Utils.jsonOption(json \ "Completion Time").map(_.extract[Long])
+    val failureReason = Utils.jsonOption(json \ "Failure Reason").map(_.extract[String])
     val emittedTaskSizeWarning = (json \ "Emitted Task Size Warning").extract[Boolean]
 
     val stageInfo = new StageInfo(stageId, stageName, numTasks, rddInfo)
     stageInfo.submissionTime = submissionTime
     stageInfo.completionTime = completionTime
+    stageInfo.failureReason = failureReason
     stageInfo.emittedTaskSizeWarning = emittedTaskSizeWarning
     stageInfo
   }
@@ -561,8 +563,7 @@ private[spark] object JsonProtocol {
       case `jobSucceeded` => JobSucceeded
       case `jobFailed` =>
         val exception = exceptionFromJson(json \ "Exception")
-        val failedStageId = (json \ "Failed Stage ID").extract[Int]
-        new JobFailed(exception, failedStageId)
+        new JobFailed(exception)
     }
   }