火花奇怪的 ORC NullPointerException

Question

调用处理所有数据的任何操作时，我在 spark 中收到一个奇怪的 NullPointerException。 有趣的是，

val dOverallTotal = spark.read.orc("/path/to/file.orc/")
dOverallTotal.distinct.count // NPE!

有趣的是，由于空指针异常而失败：

val dOverallTotal = spark.read.orc("/path/to/file.orc/partition=part_value")
dOverallTotal.distinct.count // works just fine

单独处理时对所有可用目录/part_values 是否正常工作？

下面的堆栈跟踪：

java.lang.NullPointerException
        at org.apache.spark.sql.execution.datasources.orc.OrcColumnVector.getInt(OrcColumnVector.java:132)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.agg_doAggregateWithKeys_0$(Unknown Source)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.processNext(Unknown Source)
        at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
        at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$10$$anon$1.hasNext(WholeStageCodegenExec.scala:614)
        at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
        at org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:96)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:53)
        at org.apache.spark.scheduler.Task.run(Task.scala:109)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:345)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)
20/04/06 13:12:05 WARN TaskSetManager: Lost task 9.0 in stage 138.0 (TID 976, localhost, executor driver): java.lang.NullPointerException
        at org.apache.spark.sql.execution.datasources.orc.OrcColumnVector.getInt(OrcColumnVector.java:132)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.agg_doAggregateWithKeys_0$(Unknown Source)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.processNext(Unknown Source)
        at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
        at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$10$$anon$1.hasNext(WholeStageCodegenExec.scala:614)
        at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
        at org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:96)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:53)
        at org.apache.spark.scheduler.Task.run(Task.scala:109)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:345)
        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)

20/04/06 13:12:05 ERROR TaskSetManager: Task 9 in stage 138.0 failed 1 times; aborting job
20/04/06 13:12:05 WARN TaskSetManager: Lost task 31.0 in stage 138.0 (TID 998, localhost, executor driver): TaskKilled (Stage cancelled)
20/04/06 13:12:05 WARN TaskSetManager: Lost task 19.0 in stage 138.0 (TID 986, localhost, executor driver): TaskKilled (Stage cancelled)
20/04/06 13:12:05 WARN TaskSetManager: Lost task 16.0 in stage 138.0 (TID 983, localhost, executor driver): TaskKilled (Stage cancelled)
20/04/06 13:12:05 WARN TaskSetManager: Lost task 0.0 in stage 138.0 (TID 967, localhost, executor driver): TaskKilled (Stage cancelled)
20/04/06 13:12:05 WARN TaskSetManager: Lost task 4.0 in stage 138.0 (TID 971, localhost, executor driver): TaskKilled (Stage cancelled)
nterException
        at org.apache.spark.sql.execution.datasources.orc.OrcColumnVector.getInt(OrcColumnVector.java:132)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.agg_doAggregateWithKeys_0$(Unknown Source)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.processNext(Unknown Source)
        at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
        at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$10$$anon$1.hasNext(WholeStageCodegenExec.scala:614)
        at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:409)
        at org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:125)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:96)
        at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:53)
        at org.apache.spark.scheduler.Task.run(Task.scala:109)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:345)
        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)

我的 spark 版本是 HDP 3.1 上的 2.3.x。

Answer 1

对我来说，事实证明我可以使用存储在 hive 元存储中的架构信息：

val m = spark.sessionState.catalog.getTableMetadata(TableIdentifier("table", Some("db")))
//val format = m.storage.serde.get
val format = "org.apache.hadoop.hive.ql.io.orc.OrcSerde"
spark.read
        .option("basePath", m.location.toString)
        .format(format)
        .schema(m.schema)
        .load("/input/orc/file/path/to/directory")

我想我在架构和实际存储的内容之间存在不一致。