Uploaded image for project: 'Spark'
  1. Spark
  2. SPARK-33906

SPARK UI Executors page stuck when ExecutorSummary.peakMemoryMetrics is unset

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Blocker
    • Resolution: Fixed
    • 3.2.0
    • 3.1.0
    • Web UI
    • None

    Description

      How to reproduce it?

      In mac OS standalone mode, open a spark-shell and run

      $SPARK_HOME/bin/spark-shell --master spark://localhost:7077

      val x = sc.makeRDD(1 to 100000, 5)
      x.count()
      

      Then open the app UI in the browser, and click the Executors page, will get stuck at this page: 

      Also the return JSON of REST API endpoint http://localhost:4040/api/v1/applications/app-20201224134418-0003/executors miss "peakMemoryMetrics" for executors.

      [ {
        "id" : "driver",
        "hostPort" : "192.168.1.241:50042",
        "isActive" : true,
        "rddBlocks" : 0,
        "memoryUsed" : 0,
        "diskUsed" : 0,
        "totalCores" : 0,
        "maxTasks" : 0,
        "activeTasks" : 0,
        "failedTasks" : 0,
        "completedTasks" : 0,
        "totalTasks" : 0,
        "totalDuration" : 0,
        "totalGCTime" : 0,
        "totalInputBytes" : 0,
        "totalShuffleRead" : 0,
        "totalShuffleWrite" : 0,
        "isBlacklisted" : false,
        "maxMemory" : 455501414,
        "addTime" : "2020-12-24T19:44:18.033GMT",
        "executorLogs" : { },
        "memoryMetrics" : {
          "usedOnHeapStorageMemory" : 0,
          "usedOffHeapStorageMemory" : 0,
          "totalOnHeapStorageMemory" : 455501414,
          "totalOffHeapStorageMemory" : 0
        },
        "blacklistedInStages" : [ ],
        "peakMemoryMetrics" : {
          "JVMHeapMemory" : 135021152,
          "JVMOffHeapMemory" : 149558576,
          "OnHeapExecutionMemory" : 0,
          "OffHeapExecutionMemory" : 0,
          "OnHeapStorageMemory" : 3301,
          "OffHeapStorageMemory" : 0,
          "OnHeapUnifiedMemory" : 3301,
          "OffHeapUnifiedMemory" : 0,
          "DirectPoolMemory" : 67963178,
          "MappedPoolMemory" : 0,
          "ProcessTreeJVMVMemory" : 0,
          "ProcessTreeJVMRSSMemory" : 0,
          "ProcessTreePythonVMemory" : 0,
          "ProcessTreePythonRSSMemory" : 0,
          "ProcessTreeOtherVMemory" : 0,
          "ProcessTreeOtherRSSMemory" : 0,
          "MinorGCCount" : 15,
          "MinorGCTime" : 101,
          "MajorGCCount" : 0,
          "MajorGCTime" : 0
        },
        "attributes" : { },
        "resources" : { },
        "resourceProfileId" : 0,
        "isExcluded" : false,
        "excludedInStages" : [ ]
      }, {
        "id" : "0",
        "hostPort" : "192.168.1.241:50054",
        "isActive" : true,
        "rddBlocks" : 0,
        "memoryUsed" : 0,
        "diskUsed" : 0,
        "totalCores" : 12,
        "maxTasks" : 12,
        "activeTasks" : 0,
        "failedTasks" : 0,
        "completedTasks" : 5,
        "totalTasks" : 5,
        "totalDuration" : 2107,
        "totalGCTime" : 25,
        "totalInputBytes" : 0,
        "totalShuffleRead" : 0,
        "totalShuffleWrite" : 0,
        "isBlacklisted" : false,
        "maxMemory" : 455501414,
        "addTime" : "2020-12-24T19:44:20.335GMT",
        "executorLogs" : {
          "stdout" : "http://192.168.1.241:8081/logPage/?appId=app-20201224134418-0003&executorId=0&logType=stdout",
          "stderr" : "http://192.168.1.241:8081/logPage/?appId=app-20201224134418-0003&executorId=0&logType=stderr"
        },
        "memoryMetrics" : {
          "usedOnHeapStorageMemory" : 0,
          "usedOffHeapStorageMemory" : 0,
          "totalOnHeapStorageMemory" : 455501414,
          "totalOffHeapStorageMemory" : 0
        },
        "blacklistedInStages" : [ ],
        "attributes" : { },
        "resources" : { },
        "resourceProfileId" : 0,
        "isExcluded" : false,
        "excludedInStages" : [ ]
      } ]
      

      I debugged it and observed that ExecutorMetricsPoller
      .getExecutorUpdates returns an empty map, which causes peakExecutorMetrics to None in https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/status/LiveEntity.scala#L345. The possible reason for returning the empty map is that the stage completion time is shorter than the heartbeat interval, so the stage entry in stageTCMP has already been removed before the reportHeartbeat is called.

      How to fix it?

      Check if the peakMemoryMetrics is undefined in executorspage.js.

      Attachments

        1. executor-page.png
          363 kB
          Baohe Zhang

        Issue Links

          Activity

            People

              Baohe Zhang Baohe Zhang
              Baohe Zhang Baohe Zhang
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: