Spark 2.2 空安全左外连接空指针异常

Posted

技术标签:

【中文标题】Spark 2.2 空安全左外连接空指针异常【英文标题】：Spark 2.2 Null-safe Left Outer Join Null Pointer Exception 【发布时间】：2017-10-10 14:12:31 【问题描述】：

使用 null 安全的等于运算符执行左外连接会导致 NullPointerException。

版本 火花2.2.0， 斯卡拉 2.11.8

scala> var d1 = Seq((null, 1), ("a1", 2)).toDF("a", "b")
scala> d1.show
+----+---+
|   a|  b|
+----+---+
|null|  1|
|  a1|  2|
+----+---+

scala> var d2 = Seq(("a2", 3)).toDF("a", "b")
scala> d2.show
+---+---+
|  a|  b|
+---+---+
| a2|  3|
+---+---+

scala> d1.joinWith(d2, d1("a") <=> d2("a"), "left_outer").show
17/10/10 09:44:39 ERROR Executor: Exception in task 0.0 in stage 6.0 (TID 8)
java.lang.NullPointerException

这是预期的行为吗？

堆栈跟踪

java.lang.NullPointerException
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
    at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
    at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:395)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:234)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:228)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
    at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
    at org.apache.spark.scheduler.Task.run(Task.scala:108)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:335)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)
17/10/10 10:19:28 WARN TaskSetManager: Lost task 0.0 in stage 0.0 (TID 0, localhost, executor driver): java.lang.NullPointerException
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
    at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
    at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:395)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:234)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:228)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
    at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
    at org.apache.spark.scheduler.Task.run(Task.scala:108)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:335)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)

17/10/10 10:19:28 ERROR TaskSetManager: Task 0 in stage 0.0 failed 1 times; aborting job
org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 0.0 failed 1 times, most recent failure: Lost task 0.0 in stage 0.0 (TID 0, localhost, executor driver): java.lang.NullPointerException
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
    at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
    at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:395)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:234)
    at org.apache.spark.sql.execution.SparkPlan$$anonfun$2.apply(SparkPlan.scala:228)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$25.apply(RDD.scala:827)
    at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
    at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
    at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
    at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
    at org.apache.spark.scheduler.Task.run(Task.scala:108)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:335)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)

【问题讨论】：

【参考方案1】：

我可以说在 Spark 2.2.0 和 Scala 2.11.8 版本中没有发现任何问题，因为在使用具有 null 安全的相同代码示例时我没有遇到任何异常

和等号运算符

===

您能否再次检查并添加与问题相关的更多详细信息？

val d1 = sc.parallelize(Seq(
          (null, 1), ("a1",2))
        ).toDF("a", "b") 

d1.show


+----+---+
|   a|  b|
+----+---+
|null|  1|
|  a1|  2|
+----+---+


val d2 = sc.parallelize(Seq(
      ("a2",3))
    ).toDF("a", "b") 

d2.show

+---+---+
|  a|  b|
+---+---+
| a2|  3|
+---+---+


d1.joinWith(d2, d1("a") <=>  d2("a"), "left_outer").show()

+--------+----+
|      _1|  _2|
+--------+----+
|[null,1]|null|
|  [a1,2]|null|
+--------+----+

d1.joinWith(d2, d1("a") ===  d2("a"), "left_outer").show()



+--------+----+
|      _1|  _2|
+--------+----+
|[null,1]|null|
|  [a1,2]|null|
+--------+----+

添加其他示例：

val x = Seq((100L,null), (102L,"17179869185L"), (101L,"17179869186L"), (200L,"17179869186L"), (401L,"1L"), (500L,"1L"), (600L,"8589934593L"), (700L,"8589934593L"), (800L,"8589934593L"), (900L,"8589934594L"), (1000L,"8589934594L"), (1200L,"2L"), (1300L,"2L"), (1301L,"2L"), (1400L,"17179869187L"), (1500L,"17179869188L"), (1600L,"8589934595L")).toDF("u","x1")

x.show()


+----+------------+
|   u|          x1|
+----+------------+
| 100|        null|
| 102|17179869185L|
| 101|17179869186L|
| 200|17179869186L|
| 401|          1L|
| 500|          1L|
| 600| 8589934593L|
| 700| 8589934593L|
| 800| 8589934593L|
| 900| 8589934594L|
|1000| 8589934594L|
|1200|          2L|
|1300|          2L|
|1301|          2L|
|1400|17179869187L|
|1500|17179869188L|
|1600| 8589934595L|
+----+------------+

val y = Seq(("17179869187L",-8589934595L), ("17179869188L",-8589934595L), ("17179869185L",-858993
4593L)).toDF("x2","y")


y.show()

+------------+-----------+
|          x2|          y|
+------------+-----------+
|17179869187L|-8589934595|
|17179869188L|-8589934595|
|17179869185L|-8589934593|
+------------+-----------+


x.join(y,'x1 === 'x2, "left_outer").show()

+----+------------+------------+-----------+
|   u|          x1|          x2|          y|
+----+------------+------------+-----------+
| 100|        null|        null|       null|
| 102|17179869185L|17179869185L|-8589934593|
| 101|17179869186L|        null|       null|
| 200|17179869186L|        null|       null|
| 401|          1L|        null|       null|
| 500|          1L|        null|       null|
| 600| 8589934593L|        null|       null|
| 700| 8589934593L|        null|       null|
| 800| 8589934593L|        null|       null|
| 900| 8589934594L|        null|       null|
|1000| 8589934594L|        null|       null|
|1200|          2L|        null|       null|
|1300|          2L|        null|       null|
|1301|          2L|        null|       null|
|1400|17179869187L|17179869187L|-8589934595|
|1500|17179869188L|17179869188L|-8589934595|
|1600| 8589934595L|        null|       null|
+----+------------+------------+-----------+

x: org.apache.spark.sql.DataFrame = [u: bigint, x1: string]
y: org.apache.spark.sql.DataFrame = [x2: string, y: bigint]
Command took 1.00 second

【讨论】：

我尝试了您的示例并获得了相同的输出。我给出的示例仍然给出相同的错误，也许sc.parallelize().toDF()和Seq().toDF()之间存在差异 @rdg 已经有一段时间了。你找到这个解决方案了吗？从Seq 创建Dataset 时遇到同样的问题。

以上是关于Spark 2.2 空安全左外连接空指针异常的主要内容，如果未能解决你的问题，请参考以下文章

使用 Hbase 运行 Nutch 爬虫 2.2 时出现空指针异常

Spark程序的Map函数中的空指针异常

Spark 之访问 Hive 空指针异常（Windows）

火花流中的广播变量空指针异常

解决spark dataframe get 报空指针异常 java.lang.NullPointerException

Kotlin空安全总结 ( 变量可空性 | 手动空安全管理 | 空安全调用操作符 | 非空断言操作符 | 空合并操作符 | 空指针异常处理 | 先决条件函数判空 )