[jira] [Created] (SPARK-5387) parquet writer runs into OOM during writing when number of rows is large

["Shirley Wu (JIRA)" <ji...@apache.org>]

Shirley Wu created SPARK-5387:
---------------------------------

             Summary: parquet writer runs into OOM during writing when number of rows is large
                 Key: SPARK-5387
                 URL: https://issues.apache.org/jira/browse/SPARK-5387
             Project: Spark
          Issue Type: Bug
          Components: Input/Output
    Affects Versions: 1.1.1
            Reporter: Shirley Wu


When the number of records is large in RDD, the saveAsParquet will have OOM.

Here is the stack trace:

    15/01/23 10:00:02 WARN TaskSetManager: Lost task 0.0 in stage 2.0 (TID 2, hdc2-s3.niara.com): java.lang.OutOfMemoryError: Java heap space
        parquet.bytes.CapacityByteArrayOutputStream.initSlabs(CapacityByteArrayOutputStream.java:65)
        parquet.bytes.CapacityByteArrayOutputStream.<init>(CapacityByteArrayOutputStream.java:57)
        parquet.column.values.rle.RunLengthBitPackingHybridEncoder.<init>(RunLengthBitPackingHybridEncoder.java:125)
        parquet.column.values.rle.RunLengthBitPackingHybridValuesWriter.<init>(RunLengthBitPackingHybridValuesWriter.java:36)
        parquet.column.ParquetProperties.getColumnDescriptorValuesWriter(ParquetProperties.java:61)
        parquet.column.impl.ColumnWriterImpl.<init>(ColumnWriterImpl.java:73)
        parquet.column.impl.ColumnWriteStoreImpl.newMemColumn(ColumnWriteStoreImpl.java:68)
        parquet.column.impl.ColumnWriteStoreImpl.getColumnWriter(ColumnWriteStoreImpl.java:56)
        parquet.io.MessageColumnIO$MessageColumnIORecordConsumer.<init>(MessageColumnIO.java:124)
        parquet.io.MessageColumnIO.getRecordWriter(MessageColumnIO.java:315)
        parquet.hadoop.InternalParquetRecordWriter.initStore(InternalParquetRecordWriter.java:106)
        parquet.hadoop.InternalParquetRecordWriter.checkBlockSizeReached(InternalParquetRecordWriter.java:126)
        parquet.hadoop.InternalParquetRecordWriter.write(InternalParquetRecordWriter.java:117)
        parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:81)
        parquet.hadoop.ParquetRecordWriter.write(ParquetRecordWriter.java:37)
        org.apache.spark.sql.parquet.InsertIntoParquetTable.org$apache$spark$sql$parquet$InsertIntoParquetTable$$writeShard$1(ParquetTableOperations.scala:303)
        org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:318)
        org.apache.spark.sql.parquet.InsertIntoParquetTable$$anonfun$saveAsHadoopFile$1.apply(ParquetTableOperations.scala:318)
        org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:62)
        org.apache.spark.scheduler.Task.run(Task.scala:54)
        org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:180)
        java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        java.lang.Thread.run(Thread.java:745)

It seems the writeShard() API needs to flush to disk periodically.



--
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