RapidMiner

Mahout clustering test failing on connectivity full test

SOLVED
Contributor II

Mahout clustering test failing on connectivity full test

[ Edited ]

I am running RapidMiner Studio 7.2 and connecting to a stock Cloudera QuickStart 5.7 VM with 4 CPUs and 24GB of RAM. When I run the connectivity full test, it fails on the Mahout clustering job. It looks like the error involves a java heap allocation, but I am not sure which one I should change. I have attached the test logs and added the map reduce log into the zip file. Does anyone have any idea what is going on?

 

Thanks!

Attachments

2 REPLIES
Contributor II

Re: Mahout clustering test failing on connectivity full test

Hi,

the Mahout jobs usually need more memory for the Map tasks as the Quickstart VM's default configuration. Try to set 

mapreduce.map.java.opts=-Xmx256m

as an Advanced Hadoop parameter.

Contributor II

Re: Mahout clustering test failing on connectivity full test

Thanks! That also fixed the Pig script test, which was also failing. I am completing the entire test now.