[HADOOP-2943] Compression for intermediate map output is broken

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.17.0
• Fix Version/s: 0.17.0
• Assignee: Christopher Douglas
• Reporter: Christopher Douglas
• Created: 06/Mar/08 02:30
• Updated: 08/Jul/09 16:52
• Resolved: 09/Mar/08 04:05

Description

It looks like SequenceFile::RecordCompressWriter and SequenceFile::BlockCompressWriter weren't updated to use the new serialization added in HADOOP-1986. This causes failures in the merge when mapred.compress.map.output is true and mapred.map.output.compression.type=BLOCK:

java.io.IOException: File is corrupt!
        at org.apache.hadoop.io.SequenceFile$Reader.readBlock(SequenceFile.java:1656)
        at org.apache.hadoop.io.SequenceFile$Reader.nextRawKey(SequenceFile.java:1969)
        at org.apache.hadoop.io.SequenceFile$Sorter$SegmentDescriptor.nextRawKey(SequenceFile.java:2985)
        at org.apache.hadoop.io.SequenceFile$Sorter$MergeQueue.merge(SequenceFile.java:2785)
        at org.apache.hadoop.io.SequenceFile$Sorter.merge(SequenceFile.java:2494)
        at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.mergeParts(MapTask.java:654)
        at org.apache.hadoop.mapred.MapTask$MapOutputBuffer.flush(MapTask.java:740)
        at org.apache.hadoop.mapred.MapTask.run(MapTask.java:212)
        at org.apache.hadoop.mapred.TaskTracker$Child.main(TaskTracker.java:2077)

mapred.map.output.compression.type=RECORD works for Writables, but should be updated.

Issue Links

• blocks: HADOOP-2919 Create fewer copies of buffer data during sort/spill