1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
|
/*
* 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 org.scalatest.FunSuite
import org.apache.spark.{SparkContext, LocalSparkContext}
import scala.collection.mutable
import org.scalatest.matchers.ShouldMatchers
import org.apache.spark.SparkContext._
/**
*
*/
class SparkListenerSuite extends FunSuite with LocalSparkContext with ShouldMatchers {
// TODO: This test has a race condition since the DAGScheduler now reports results
// asynchronously. It needs to be updated for that patch.
ignore("local metrics") {
sc = new SparkContext("local[4]", "test")
val listener = new SaveStageInfo
sc.addSparkListener(listener)
sc.addSparkListener(new StatsReportListener)
//just to make sure some of the tasks take a noticeable amount of time
val w = {i:Int =>
if (i == 0)
Thread.sleep(100)
i
}
val d = sc.parallelize(1 to 1e4.toInt, 64).map{i => w(i)}
d.count
Thread.sleep(1000)
listener.stageInfos.size should be (1)
val d2 = d.map{i => w(i) -> i * 2}.setName("shuffle input 1")
val d3 = d.map{i => w(i) -> (0 to (i % 5))}.setName("shuffle input 2")
val d4 = d2.cogroup(d3, 64).map{case(k,(v1,v2)) => w(k) -> (v1.size, v2.size)}
d4.setName("A Cogroup")
d4.collectAsMap
Thread.sleep(1000)
listener.stageInfos.size should be (4)
listener.stageInfos.foreach {stageInfo =>
//small test, so some tasks might take less than 1 millisecond, but average should be greater than 1 ms
checkNonZeroAvg(stageInfo.taskInfos.map{_._1.duration}, stageInfo + " duration")
checkNonZeroAvg(stageInfo.taskInfos.map{_._2.executorRunTime.toLong}, stageInfo + " executorRunTime")
checkNonZeroAvg(stageInfo.taskInfos.map{_._2.executorDeserializeTime.toLong}, stageInfo + " executorDeserializeTime")
if (stageInfo.stage.rdd.name == d4.name) {
checkNonZeroAvg(stageInfo.taskInfos.map{_._2.shuffleReadMetrics.get.fetchWaitTime}, stageInfo + " fetchWaitTime")
}
stageInfo.taskInfos.foreach{case (taskInfo, taskMetrics) =>
taskMetrics.resultSize should be > (0l)
if (isStage(stageInfo, Set(d2.name, d3.name), Set(d4.name))) {
taskMetrics.shuffleWriteMetrics should be ('defined)
taskMetrics.shuffleWriteMetrics.get.shuffleBytesWritten should be > (0l)
}
if (stageInfo.stage.rdd.name == d4.name) {
taskMetrics.shuffleReadMetrics should be ('defined)
val sm = taskMetrics.shuffleReadMetrics.get
sm.totalBlocksFetched should be > (0)
sm.localBlocksFetched should be > (0)
sm.remoteBlocksFetched should be (0)
sm.remoteBytesRead should be (0l)
sm.remoteFetchTime should be (0l)
}
}
}
}
def checkNonZeroAvg(m: Traversable[Long], msg: String) {
assert(m.sum / m.size.toDouble > 0.0, msg)
}
def isStage(stageInfo: StageInfo, rddNames: Set[String], excludedNames: Set[String]) = {
val names = Set(stageInfo.stage.rdd.name) ++ stageInfo.stage.rdd.dependencies.map{_.rdd.name}
!names.intersect(rddNames).isEmpty && names.intersect(excludedNames).isEmpty
}
class SaveStageInfo extends SparkListener {
val stageInfos = mutable.Buffer[StageInfo]()
override def onStageCompleted(stage: StageCompleted) {
stageInfos += stage.stageInfo
}
}
}
|