spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Cheng Lian (JIRA)" <j...@apache.org>
Subject [jira] [Created] (SPARK-6360) For Spark 1.2, after any RDD transformations, calling saveAsParquetFile over a SchemaRDD with decimal column throws
Date Mon, 16 Mar 2015 16:19:38 GMT
Cheng Lian created SPARK-6360:
---------------------------------

             Summary: For Spark 1.2, after any RDD transformations, calling saveAsParquetFile
over a SchemaRDD with decimal column throws
                 Key: SPARK-6360
                 URL: https://issues.apache.org/jira/browse/SPARK-6360
             Project: Spark
          Issue Type: Bug
          Components: SQL
    Affects Versions: 1.2.0, 1.1.0
            Reporter: Cheng Lian


Spark shell session for reproduction (use {{:paste}}):
{noformat}
import org.apache.spark.sql.SQLContext
import org.apache.spark.sql.catalyst.types.decimal._
import org.apache.spark.sql.catalyst.types._
import org.apache.hadoop.fs._

val sqlContext = new SQLContext(sc)
val fs = FileSystem.get(sc.hadoopConfiguration)

fs.delete(new Path("a.parquet"))
fs.delete(new Path("b.parquet"))

import sc._
import sqlContext._

val r1 = parallelize(1 to 10)
  .map(i => Tuple1(Decimal(i, 10, 0)))
  .select('_1 cast DecimalType(10, 0))

// OK
r1.saveAsParquetFile("a.parquet")

val r2 = parallelize(1 to 10)
  .map(i => Tuple1(Decimal(i, 10, 0)))
  .select('_1 cast DecimalType(10, 0))

val r3 = r2.coalesce(1)

// Error
r3.saveAsParquetFile("b.parquet")
{noformat}
Exception thrown:
{noformat}
java.lang.ClassCastException: scala.math.BigDecimal cannot be cast to org.apache.spark.sql.catalyst.types.decimal.Decimal
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.consumeType(ParquetTableSupport.scala:359)
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.write(ParquetTableSupport.scala:328)
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.write(ParquetTableSupport.scala:314)
        at parquet.hadoop.InternalParquetRecordWriter.write(InternalParquetRecordWriter.java:120)
        at parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:81)
        at parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:37)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable.org$apache$spark$sql$parquet$InsertIntoParquetTable$$writeShard$1(ParquetTableOperations.scala:308)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:325)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:325)
        at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:61)
        at org.apache.spark.scheduler.Task.run(Task.scala:56)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:196)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
        at java.lang.Thread.run(Thread.java:745)
15/03/17 00:04:13 WARN TaskSetManager: Lost task 0.0 in stage 2.0 (TID 2, localhost): java.lang.ClassCastException:
scala.math.BigDecimal cannot be cast to org.apache.spark.sql.catalyst.types.decimal.Decimal
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.consumeType(ParquetTableSupport.scala:359)
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.write(ParquetTableSupport.scala:328)
        at org.apache.spark.sql.parquet.MutableRowWriteSupport.write(ParquetTableSupport.scala:314)
        at parquet.hadoop.InternalParquetRecordWriter.write(InternalParquetRecordWriter.java:120)
        at parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:81)
        at parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:37)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable.org$apache$spark$sql$parquet$InsertIntoParquetTable$$writeShard$1(ParquetTableOperations.scala:308)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:325)
        at org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:325)
        at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:61)
        at org.apache.spark.scheduler.Task.run(Task.scala:56)
        at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:196)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
        at java.lang.Thread.run(Thread.java:745)
{noformat}
The query plan of {{r1}} is:
{noformat}
== Parsed Logical Plan ==
'Project [CAST('_1, DecimalType(10,0)) AS c0#60]
 LogicalRDD [_1#59], MapPartitionsRDD[71] at mapPartitions at ExistingRDD.scala:36

== Analyzed Logical Plan ==
Project [CAST(_1#59, DecimalType(10,0)) AS c0#60]
 LogicalRDD [_1#59], MapPartitionsRDD[71] at mapPartitions at ExistingRDD.scala:36

== Optimized Logical Plan ==
Project [CAST(_1#59, DecimalType(10,0)) AS c0#60]
 LogicalRDD [_1#59], MapPartitionsRDD[71] at mapPartitions at ExistingRDD.scala:36

== Physical Plan ==
Project [CAST(_1#59, DecimalType(10,0)) AS c0#60]
 PhysicalRDD [_1#59], MapPartitionsRDD[71] at mapPartitions at ExistingRDD.scala:36

Code Generation: false
== RDD ==
{noformat}
while {{r3}}'s query plan is:
{noformat}
== Parsed Logical Plan ==
LogicalRDD [c0#61], CoalescedRDD[74] at coalesce at SchemaRDD.scala:456

== Analyzed Logical Plan ==
LogicalRDD [c0#61], CoalescedRDD[74] at coalesce at SchemaRDD.scala:456

== Optimized Logical Plan ==
LogicalRDD [c0#61], CoalescedRDD[74] at coalesce at SchemaRDD.scala:456

== Physical Plan ==
PhysicalRDD [c0#61], CoalescedRDD[74] at coalesce at SchemaRDD.scala:456

Code Generation: false
== RDD ==
{noformat}
The key difference here is that, {{r3}} wraps an existing {{SchemaRDD}} ({{r2}}, beneath the
{{CoalescedRDD}}). While evaluating {{r3}}, {{r2.compute}} is called, which calls {{ScalaReflection.convertRowToScala}}.
Here, Catalyst {{Decimal}} values are converted into Java {{BigDecimals}}, and finally causes
the exception.

Note that {{DataFrame}} in Spark 1.3 doesn't suffer this issue.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message