simple test w/ failure involving a shared dependency

core/src/test/scala/org/apache/spark/scheduler/DAGSchedulerSuite.scala

@@ -556,11 +556,17 @@ class DAGSchedulerSuite
    * @param stageId - The current stageId
    * @param attemptIdx - The current attempt count
    */
-  private def completeNextResultStageWithSuccess(stageId: Int, attemptIdx: Int): Unit = {
+  private def completeNextResultStageWithSuccess(
+      stageId: Int,
+      attemptIdx: Int,
+      partitionToResult: Int => Int = _ => 42): Unit = {
     val stageAttempt = taskSets.last
     checkStageId(stageId, attemptIdx, stageAttempt)
     assert(scheduler.stageIdToStage(stageId).isInstanceOf[ResultStage])
-    complete(stageAttempt, stageAttempt.tasks.zipWithIndex.map(_ => (Success, 42)).toSeq)
+    val taskResults = stageAttempt.tasks.zipWithIndex.map { case (task, idx) =>
+      (Success, partitionToResult(idx))
+    }
+    complete(stageAttempt, taskResults.toSeq)
   }
 
   /**
@@ -1017,6 +1023,47 @@ class DAGSchedulerSuite
     assertDataStructuresEmpty()
   }
 
+  /**
+   * Run two jobs, with a shared dependency.  We simulate a fetch failure in the second job, which
+   * requires regenerating some outputs of the shared dependency.  One key aspect of this test is
+   * that the second job actually uses a different stage for the shared dependency (a "skipped"
+   * stage).
+   */
+  test("shuffle fetch failure in a reused shuffle dependency") {
+    // Run the first job successfully, which creates one shuffle dependency
+
+    val shuffleMapRdd = new MyRDD(sc, 2, Nil)
+    val shuffleDep = new ShuffleDependency(shuffleMapRdd, null)
+    val reduceRdd = new MyRDD(sc, 2, List(shuffleDep))
+    submit(reduceRdd, Array(0, 1))
+
+    completeShuffleMapStageSuccessfully(0, 0, 2)
+    completeNextResultStageWithSuccess(1, 0)
+    assert(results === Map(0 -> 42, 1 -> 42))
+    assertDataStructuresEmpty()
+
+    // submit another job w/ the shared dependency, and have a fetch failure
+    val reduce2 = new MyRDD(sc, 2, List(shuffleDep))
+    submit(reduce2, Array(0, 1))
+    // Note that the stage numbering here is only b/c the shared dependency produces a new, skipped
+    // stage.  If instead it reused the existing stage, then this would be stage 2
+    completeNextStageWithFetchFailure(3, 0, shuffleDep)
+    scheduler.resubmitFailedStages()
+
+    // the scheduler now creates a new task set to regenerate the missing map output, but this time
+    // using a different stage, the "skipped" one
+
+    // SPARK-9809 -- this stage is submitted without a task for each partition (because some of
+    // the shuffle map output is still available from stage 0); make sure we've still got internal
+    // accumulators setup
+    assert(scheduler.stageIdToStage(2).internalAccumulators.nonEmpty)
+    completeShuffleMapStageSuccessfully(2, 0, 2)
+    completeNextResultStageWithSuccess(3, 1, idx => idx + 1234)
+    assert(results === Map(0 -> 1234, 1 -> 1235))
+
+    assertDataStructuresEmpty()
+  }
+
   /**
    * Makes sure that failures of stage used by multiple jobs are correctly handled.
    *