Reimplementing some of the hadoop-ec2 tools without so much shell script as a way to get cozy with the AWS SDK.

Building - it's currently setup as a Netbeans project, will be getting the Netbeans stuff out soon.

depends on: --aws-sdk http://aws.amazon.com/sdkforjava/

--commons-codec v. 1.4 (aws includes 1.3 but I had issues using it) http://commons.apache.org/codec/

--log4j http://logging.apache.org/log4j/

--ganymed (ssh lib) http://www.cleondris.ch/opensource/ssh2/

setup the required libraries and build.

Configuring -- there's sample config files in dist/config to launch a cluster you need to choose or build an AMI, preferably with hadoop installed, but you could install hadoop on startup without too much effort.

Running --

run pooper.sh in dist/bin after building.

As of this writing, HaddooperPooper 's command-list: command create-groups delete-cluster describe-cluster get-login-command launch-cluster launch-master launch-slaves list-clusters login push-file terminate-cluster terminate-slaves

Still to come: port forwarding, automated job submission/termination, support for a map-reduce only (non-dfs) slave group that can be scaled separately, GUI.