LbaaS agent scheduling error exposes implementation

[Eugene Nikanorov]

Right now when there is no active lbaas agent for HAProxy driver (or other agent-based driver), the following error is returned to the client:

"No eligible loadbalancer agent found for pool <pool_id>"

We need to return some more generic error to the user, skipping the notion of agent.

Also, pool will remain in PENDING_CREATE state, while it probably should move to an ERROR state.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/81883

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/81883
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=98b3f4a95104234e5106bc1ee58efd427da96c00
Submitter: Jenkins
Branch: master

commit 98b3f4a95104234e5106bc1ee58efd427da96c00
Author: Eugene Nikanorov <email address hidden>
Date: Thu Mar 20 18:43:18 2014 +0400

    Return meaningful error message on pool creation error

    Instead of returning exception specific to haproxy (agent-based),
    return more generic BackendNotFound exception.
    It also could be used later when binding to devices will be
    introduced. That exception will indicate scheduling failure, e.g.
    inability to find appropriate backend for the resource.
    Resource (pool) is then marked with ERROR state with corresponding
    error description.

    Change-Id: Ic18ff20102b4bb2b97e7b186fcf797133bd3ba3d
    Closes-Bug: #1295214