mapreduce on CDH5: not enough memory for Yarn about spring-hadoop-samples

[trisberg]

Hi, thanks for trying this. I know I tested the sample on the CDH5 Quickstart VM and don't think I changed that setting. Did you use the Quickstart VM or did you use your own install?

[renaud]

Thanks Thomas for writing back.
I used the Quickstart VM.

[trisberg]

Strange. Looking at my config -

Container Memory Minimum 1 GiB default value
yarn.scheduler.minimum-allocation-mb

So my mapreduce example runs in less than 1GB.

I assume you have the latest with the Spring for Apache Hadoop 2.0 and build using the cdh5 profile:

mvn clean package -Pcdh5
sh target/appassembler/bin/wordcount

[renaud]

Yes, I did build with the cdh5 profile. So if you can't reproduce let's close this issue. It's working on my end, so all good.
Thanks, Renaud

[trisberg]

I just tried a new install of the Quickstart VM and I'm able to reproduce this - must have made some config change on the other VM I tried - I'll add a note in the instructions.

[trisberg]

Thanks for reporting this.

I have updated the README

[renaud]

Ok. Thanks too for the prompt reply!