Fix exception handling.

[jarovo]

The subcommand of the action was called using os.system. The exit value
of os.system is a 16-bit value. This value was propagated and used as
exit value of the whole cluster {ACTION} command without any
modification, resulting in exit() being called with value > 255. In
the CPython 2.7 exit(v) with v > 255 behaves like exit(0), which hides
that we had an error during the execution.

This commit removes the error propagation by return value and introduces
using exceptions instead.

[openshift-bot]

Can one of the admins verify this patch?

[jarovo]

Bump!

[abutcher]

👍 from me - @sdodson @detiber ?

[sdodson]

LGTM

[sdodson]

ok to test
no idea if the bot listens to me or not

[jarovo]

To reproduce the error:

Python 2.7.10 (default, Sep 24 2015, 17:50:09) 
[GCC 5.1.1 20150618 (Red Hat 5.1.1-4)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import os
>>> r=os.system('exit 30')
>>> print r
7680
>>> exit(r)

$ echo $?
0

[detiber]

@twiest @wshearn

[twiest]

ok to test

[twiest]

👍

@JaryN thanks for the commit! :)