[PIG-3208] [zebra] TFile should not set io.compression.codec.lzo.buffersize

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.12.0
• Patch Info: Patch Available
• Hadoop Flags: Reviewed
• Assignee: Eugene Joseph Koontz
• Reporter: Eugene Joseph Koontz
• Created: 22/Feb/13 02:30
• Updated: 14/Oct/13 16:46
• Resolved: 19/Mar/13 23:34

Description

In contrib/zebra/src/java/org/apache/hadoop/zebra/tfile/Compression.java, the following occurs:

conf.setInt("io.compression.codec.lzo.buffersize", 64 * 1024);

This can cause the LZO decompressor, if called within the context of reading TFiles, to return with an error code when trying to uncompress LZO-compressed data, if the data's compressed size is too large to fit in 64 * 1024 bytes.

For example, the Hadoop-LZO code uses a different default value (256 * 1024):

https://github.com/twitter/hadoop-lzo/blob/master/src/java/com/hadoop/compression/lzo/LzoCodec.java#L185

This can lead to a case where, if data is compressed with a cluster where the default io.compression.codec.lzo.buffersize = 256*1024 is used, then code that tries to read this data by using Pig's zebra, the Mapper will exit with code 134 because the LZO compressor returns a -4 (which encodes the LZO C library error LZO_E_INPUT_OVERRUN) when trying to uncompress the data. The stack trace of such a case is shown below:

2013-02-17 14:47:50,709 INFO com.hadoop.compression.lzo.LzoCodec: Creating stream for compressor: com.hadoop.compression.lzo.LzoCompressor@6818c458 with bufferSize: 262144
2013-02-17 14:47:50,849 INFO org.apache.hadoop.io.compress.CodecPool: Paying back codec: com.hadoop.compression.lzo.LzoCompressor@6818c458
2013-02-17 14:47:50,849 INFO org.apache.hadoop.mapred.MapTask: Finished spill 3
2013-02-17 14:47:50,857 INFO org.apache.hadoop.io.compress.CodecPool: Borrowing codec: com.hadoop.compression.lzo.LzoCompressor@6818c458
2013-02-17 14:47:50,866 INFO com.hadoop.compression.lzo.LzoCodec: Creating stream for compressor: com.hadoop.compression.lzo.LzoCompressor@6818c458 with bufferSize: 262144
2013-02-17 14:47:50,879 INFO org.apache.hadoop.io.compress.CodecPool: Paying back codec: com.hadoop.compression.lzo.LzoCompressor@6818c458
2013-02-17 14:47:50,879 INFO org.apache.hadoop.mapred.MapTask: Finished spill 4
2013-02-17 14:47:50,887 INFO org.apache.hadoop.mapred.Merger: Merging 5 sorted segments
2013-02-17 14:47:50,890 INFO org.apache.hadoop.io.compress.CodecPool: Borrowing codec: com.hadoop.compression.lzo.LzoDecompressor@66a23610
2013-02-17 14:47:50,891 INFO com.hadoop.compression.lzo.LzoDecompressor: calling decompressBytesDirect with buffer with: position: 0 and limit: 262144
2013-02-17 14:47:50,891 INFO com.hadoop.compression.lzo.LzoDecompressor: read: 245688 bytes from decompressor.
2013-02-17 14:47:50,891 INFO org.apache.hadoop.io.compress.CodecPool: Borrowing codec: com.hadoop.compression.lzo.LzoDecompressor@43684706
2013-02-17 14:47:50,892 INFO com.hadoop.compression.lzo.LzoDecompressor: calling decompressBytesDirect with buffer with: position: 0 and limit: 65536
2013-02-17 14:47:50,895 INFO org.apache.hadoop.mapred.TaskLogsTruncater: Initializing logs' truncater with mapRetainSize=-1 and reduceRetainSize=-1
2013-02-17 14:47:50,897 FATAL org.apache.hadoop.mapred.Child: Error running child : java.lang.InternalError: lzo1x_decompress returned: -4
        at com.hadoop.compression.lzo.LzoDecompressor.decompressBytesDirect(Native Method)
        at com.hadoop.compression.lzo.LzoDecompressor.decompress(LzoDecompressor.java:307)
        at org.apache.hadoop.io.compress.BlockDecompressorStream.decompress(BlockDecompressorStream.java:82)
        at org.apache.hadoop.io.compress.DecompressorStream.read(DecompressorStream.java:75)
        at org.apache.hadoop.mapred.IFile$Reader.readData(IFile.java:341)
        at org.apache.hadoop.mapred.IFile$Reader.rejigData(IFile.java:371)
        at org.apache.hadoop.mapred.IFile$Reader.readNextBlock(IFile.java:355)
        at org.apache.hadoop.mapred.IFile$Reader.next(IFile.java:387)
        at org.apache.hadoop.mapred.Merger$Segment.next(Merger.java:220)
        at org.apache.hadoop.mapred.Merger$MergeQueue.merge(Merger.java:420)
        at org.apache.hadoop.mapred.Merger$MergeQueue.merge(Merger.java:381)
        at org.apache.hadoop.mapred.Merger.merge(Merger.java:77)
        at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.mergeParts(MapTask.java:1548)
        at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.flush(MapTask.java:1180)
        at org.apache.hadoop.mapred.MapTask$NewOutputCollector.close(MapTask.java:582)
        at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:649)
        at org.apache.hadoop.mapred.MapTask.run(MapTask.java:323)
        at org.apache.hadoop.mapred.Child$4.run(Child.java:270)
        at java.security.AccessController.doPrivileged(Native Method)
        at javax.security.auth.Subject.doAs(Subject.java:396)
        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1213)
        at org.apache.hadoop.mapred.Child.main(Child.java:264)

(Some additional LOG.info statements were added to produce the above output which are not in this patch).