You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@whirr.apache.org by Sebastian Schoenherr <se...@uibk.ac.at> on 2011/03/14 13:57:49 UTC

OutOfMemoryError, mapred-site option

Hi guys,
I tried to start a 64 bit image with WHIRR (branch 0.4.0 or trunk) on 
EC2 and get a OutOfMemoryError in the child processes. It reminds me of 
this: https://issues.apache.org/jira/browse/WHIRR-146. I get the same 
distcp behavior;
After setting the mapred.child.ulimit via the property file to null 
(hadoop-mapreduce.mapred.child.ulimit=null) the distcp runs as it 
should. Maybe the option is set on a too small value or it is not 
necessary as the patch of 146 suggest.
Can anyone else reproduce this error or is it a problem of my custom image?

Cheers
Sebastian