Description of problem:

There is an inconsistency in behavior between a client-initiated yum install and server-initiated install of packages when one or more of the packages doesn't exist.

For example, (assuming unknownpkg doesn't exist)

action: client-initiated "yum install unknownpkg xterm"
result: xterm is installed

action: server (e.g. katello) initiated install of unknownpkg xterm
result: No package(s) available to install

From a user-perspective, I'd expect the behavior to be the same.  Based on discussion with Jeff, perhaps this could be allowed by introducing a new option on the server-initiated interface to support a 'best effort' install.  

Alternatively, if it cannot be supported the error message should be updated to be more accurate, since in this case xterm is available, but unknownpkg is not.

Version-Release number of selected component (if applicable):
pulp-server-2.0.6-0.2.beta.noarch

How reproducible:
always

Additional info:

The following is the complete result from pulp when attempting the server initiated install:

{"trace"=>"Traceback (most recent call last):\n\n  File \"/usr/lib/python2.7/site-packages/pulp/agent/lib/dispatcher.py\", line 75, in install\n    r = handler.install(conduit, units, dict(options))\n\n  File \"/usr/lib/pulp/agent/handlers/rpm.py\", line 99, in install\n    details = pkg.install(names)\n\n  File \"/usr/lib/python2.7/site-packages/pulp_rpm/handler/rpmtools.py\", line 124, in install\n    yb.install(pattern=info)\n\n  File \"/usr/lib/python2.7/site-packages/yum/__init__.py\", line 4157, in install\n    raise Errors.InstallError, _('No package(s) available to install')\n\nInstallError: No package(s) available to install\n", "message"=>"No package(s) available to install"}