Fixed repl maven build to produce artifacts with the appropriate hadoop classifier and extracted repl fat-jar and debian packaging into a separate project to make Maven happy

Add try-finally to handle MapOutputTracker timeouts

Have standalone cluster report exit codes to clients

Fixed the broken Java unit test from SPARK-635.

Executor exit status codes

Use spark-env.sh to configure standalone master

Also fixed a typo in the standalone mode documentation.

SPARK-635: Pass a TaskContext object to compute() interface and use that to close Hadoop input stream.

that to close Hadoop input stream. Incorporated Matei's command.

to close Hadoop input stream.

Mark hadoop dependencies provided in all library artifacts

SPARK-626: Remove security groups properly

when deleting security groups fails, fixing bug when using all zones but
only 1 slave.

Suppress shuffle block updates when a slave node comes back.

between so wait for AWS eventual consistency to catch up.

Bump CDH version for the Hadoop 2 profile to 4.1.2

Shaded repl jar for hadoop1 profile needs to include hadoop classes

SPARK-539 #resolve

Implement block manager heartbeat

Minor tweaks to the debian build

This will make it easier to make the "run" script work with a Maven build

Adds required ActorSystem argument to BlockManager constructors.