[
https://issues.apache.org/jira/browse/MAPREDUCE-5893?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14000652#comment-14000652
]
Xiangrui Meng commented on MAPREDUCE-5893:
------------------------------------------
Checked the code in the trunk. This class has a static boolean member
`skipDecompression`, which indicates whether it is decompressing or checking
the next marker.
> CBZip2InputStream is not threadsafe
> -----------------------------------
>
> Key: MAPREDUCE-5893
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5893
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: mrv1, mrv2
> Affects Versions: 1.2.1, 2.2.0
> Reporter: Xiangrui Meng
>
> Hadoop uses CBZip2InputStream to decode bzip2 files. However, the
> implementation is not threadsafe. This is not a really problem for Hadoop
> MapReduce because Hadoop runs each task in a separate JVM. But for other
> libraries that utilize multithreading and use Hadoop's InputFormat, e.g.,
> Spark, it will cause exceptions like the following:
> {code}
> java.lang.ArrayIndexOutOfBoundsException: 6
> org.apache.hadoop.io.compress.bzip2.CBZip2InputStream.recvDecodingTables(CBZip2InputStream.java:729)
>
> org.apache.hadoop.io.compress.bzip2.CBZip2InputStream.getAndMoveToFrontDecode(CBZip2InputStream.java:795)
>
> org.apache.hadoop.io.compress.bzip2.CBZip2InputStream.initBlock(CBZip2InputStream.java:499)
>
> org.apache.hadoop.io.compress.bzip2.CBZip2InputStream.changeStateToProcessABlock(CBZip2InputStream.java:330)
>
> org.apache.hadoop.io.compress.bzip2.CBZip2InputStream.read(CBZip2InputStream.java:394)
>
> org.apache.hadoop.io.compress.BZip2Codec$BZip2CompressionInputStream.read(BZip2Codec.java:428)
> java.io.InputStream.read(InputStream.java:101)
> org.apache.hadoop.util.LineReader.readDefaultLine(LineReader.java:205)
> org.apache.hadoop.util.LineReader.readLine(LineReader.java:169)
> org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:176)
> org.apache.hadoop.mapred.LineRecordReader.next(LineRecordReader.java:43)
> org.apache.spark.rdd.HadoopRDD$$anon$1.getNext(HadoopRDD.scala:198)
> org.apache.spark.rdd.HadoopRDD$$anon$1.getNext(HadoopRDD.scala:181)
> org.apache.spark.util.NextIterator.hasNext(NextIterator.scala:71)
> org.apache.spark.InterruptibleIterator.hasNext(InterruptibleIterator.scala:35)
> scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:327)
> org.apache.spark.util.Utils$.getIteratorSize(Utils.scala:1000)
> org.apache.spark.rdd.RDD$$anonfun$count$1.apply(RDD.scala:847)
> org.apache.spark.rdd.RDD$$anonfun$count$1.apply(RDD.scala:847)
> org.apache.spark.SparkContext$$anonfun$runJob$4.apply(SparkContext.scala:1077)
>
> org.apache.spark.SparkContext$$anonfun$runJob$4.apply(SparkContext.scala:1077)
> org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:111)
> org.apache.spark.scheduler.Task.run(Task.scala:51)
> org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:187)
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
>
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
> java.lang.Thread.run(Thread.java:724)
> {code}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
