hadoop streaming tasks hang for when stream.non.zero.exit.is.failure==true and 
reduce processes exit with non zero status
-------------------------------------------------------------------------------------------------------------------------

                 Key: HADOOP-3068
                 URL: https://issues.apache.org/jira/browse/HADOOP-3068
             Project: Hadoop Core
          Issue Type: Bug
          Components: contrib/streaming
         Environment: Java(TM) SE Runtime Environment (build 1.6.0_04-b12); 
Hadoop version 0.17.0-dev, r639662
            Reporter: Yuri Pradkin
             Fix For: 0.17.0


When I set *stream.non.zero.exit.is.failure* to true and run a streaming job 
with reducers that exit with a non-zero status, those tasks fail apparently 
waiting for something.

...
2008-03-21 13:33:53,715 INFO org.apache.hadoop.streaming.PipeMapRed: 
R/W/S=65501/1/0 in:334=65501/196 [rec/s] out:0=1/196 [rec/s]
2008-03-21 13:33:53,719 INFO org.apache.hadoop.streaming.PipeMapRed: 
mapRedFinished
2008-03-21 13:34:11,228 INFO org.apache.hadoop.streaming.PipeMapRed: Records 
R/W=65536/2
2008-03-21 13:34:11,235 INFO org.apache.hadoop.streaming.PipeMapRed: 
PipeMapRed.waitOutputThreads(): subprocess exitted with code 1
2008-03-21 13:34:11,235 INFO org.apache.hadoop.streaming.PipeMapRed: 
MRErrorThread done
2008-03-21 13:34:11,238 INFO org.apache.hadoop.streaming.PipeMapRed: 
MROutputThread done
2008-03-21 13:34:11,245 WARN org.apache.hadoop.mapred.TaskTracker: Error 
running child
java.lang.RuntimeException: PipeMapRed.waitOutputThreads(): subprocess failed 
with code 1
        at 
org.apache.hadoop.streaming.PipeMapRed.waitOutputThreads(PipeMapRed.java:331)
        at 
org.apache.hadoop.streaming.PipeMapRed.mapRedFinished(PipeMapRed.java:475)
        at org.apache.hadoop.streaming.PipeReducer.close(PipeReducer.java:110)
        at org.apache.hadoop.mapred.ReduceTask.run(ReduceTask.java:399)
        at 
org.apache.hadoop.mapred.TaskTracker$Child.main(TaskTracker.java:2113)

After that the task still shows up with status:Running, but it just hangs there 
and when/if all tasks get into this state, the whole cluster hangs.

BTW, may I suggest that we make *stream.non.zero.exit.is.failure* default to 
true after this is fixed?

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to