[MAPREDUCE-3388] Streaming task with special char gets wrong output

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Invalid
• Affects Version/s: 0.23.0, 2.0.0-alpha
• Component/s: contrib/streaming, mrv2
• Target Version/s: 0.23.1
• Assignee: Robert Joseph Evans
• Reporter: Robert Joseph Evans
• Created: 09/Nov/11 16:24
• Updated: 10/Mar/15 04:32
• Resolved: 08/Dec/11 22:13

Description

In 0.20.204:

hadoop jar $HADOOP_HOME/hadoop-streaming.jar '-Dmapreduce.job.acl-view-job=*' -files 'file:///tmp/InputDir#testlink!@$&*()-_+=' -input input.txt -mapper 'xargs cat' -reducer cat -output output -jobconf mapred.job.name=streamingTest-1000 -jobconf 'mapreduce.job.acl-view-job=*'

Output:

hadoop fs -cat output/*
and-so-is-the-c#
c++-also-supports-object-oriented-concepts
hadoop-apache-org-core  
hadoop-streaming
hod-is-the-part-of-hadoop
it-is-residing-on-apache-under-repos-asf
java-an object-oriented-language
smalltalk-is-also-object-oriented-language
streaming-is-also-part-of-it

For 0.23:

hadoop jar $HADOOP_MAPRED_HOME/hadoop-streaming.jar -Dmapreduce.job.acl-view-job=* -files 'file:///tmp/InputDir#testlink!@$&*()-_+=' -input input.txt  -mapper 'xargs cat' -reducer cat -output output -jobconf mapred.job.name=streamingTest-1000 -jobconf 'mapreduce.job.acl-view-job=*'

Output:

testlink!@$&*()-_+=/input1.txt testlink!@$&*()-_+=/input2.txt

The contents of input.txt are as follows:

hadoop fs -cat Streaming/streaming-1000/input.txt
testlink!@$&*()-_+=/input1.txt
testlink!@$&*()-_+=/input2.txt