I have a job with  file strd.txt is 1.5. gb with 2,458,220  records. I get
24 maps assigned for the first job.

The first 9 maps are working on the same file, with each split working on ~
333,849 records. Out of which Map m_00000 to map0006 shows complete taking
<= 6 seconds.



However the last task, map m_0007 hangs/fails.  Also m_0008 through m_00023
are shown as complete.

How to interpret this?



Also job m_007 fails due to Java Heap Space error, but this error is
eventually overwritten with timeout error , masking the real error.

In the pseudo cluster I never saw the Java Heap Space error, it always
showed me timeout error.

I am on cdh3u2.

Thanks, Prashant.

Reply via email to