diff --git a/core/src/test/scala/org/apache/spark/scheduler/ReplayListenerSuite.scala b/core/src/test/scala/org/apache/spark/scheduler/ReplayListenerSuite.scala
index e79613749f0ce..ec9bb5a338eff 100644
--- a/core/src/test/scala/org/apache/spark/scheduler/ReplayListenerSuite.scala
+++ b/core/src/test/scala/org/apache/spark/scheduler/ReplayListenerSuite.scala
@@ -22,6 +22,7 @@ import java.net.URI
 import java.util.concurrent.atomic.AtomicInteger
 
 import org.apache.hadoop.fs.Path
+import org.json4s.DefaultFormats
 import org.json4s.JsonAST.JValue
 import org.json4s.jackson.JsonMethods._
 import org.scalatest.BeforeAndAfter
@@ -218,8 +219,17 @@ class ReplayListenerSuite extends SparkFunSuite with BeforeAndAfter with LocalSp
 
     // Verify the same events are replayed in the same order
     assert(sc.eventLogger.isDefined)
-    val originalEvents = sc.eventLogger.get.loggedEvents
-    val replayedEvents = eventMonster.loggedEvents
+    implicit val format = DefaultFormats
+    def exceptCase(a: JValue) = ( a \ "Event").extract[String] match {
+      // If we are logging stage executor metrics, there is a bulk call to logEvent with
+      // SparkListenerStageExecutorMetrics events via a Map.foreach. The Map.foreach bulk
+      // operation may not log the events with the same order. So here we should not compare
+      // SparkListenerStageExecutorMetrics here.
+      case "SparkListenerStageExecutorMetrics" => false
+      case _ => true
+    }
+    val originalEvents = sc.eventLogger.get.loggedEvents.filter(exceptCase(_))
+    val replayedEvents = eventMonster.loggedEvents.filter(exceptCase(_))
     originalEvents.zip(replayedEvents).foreach { case (e1, e2) =>
       // Don't compare the JSON here because accumulators in StageInfo may be out of order
       JsonProtocolSuite.assertEquals(