gluster_native create_snapshot breaks on error

[Csaba Henk]

- On event: create_snapshot is called with gluster native driver with a setup where the backing volume does not support snapshots.
- We expect: produce a gracious error message.
- What happens: breakage with KeyError.
- Reason: glusterfs_versions dict is queried with wrongly constructed key.

Note that this was found through code audit. We'll post a bactrace as soon as we can produce the issue.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/204270

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/204270
Committed: https://git.openstack.org/cgit/openstack/manila/commit/?id=26ff88ca9f63ea24374d027252ee7501fe36451a
Submitter: Jenkins
Branch: master

commit 26ff88ca9f63ea24374d027252ee7501fe36451a
Author: Csaba Henk <email address hidden>
Date: Tue Jul 21 23:27:19 2015 +0200

    GlusterFS: fix retrieval of management address of GlusterFS volumes

    We need the so-called management address of a GlusterFS
    volume (ssh address of its management node) at a few places
    for bookkeping reasons (certain data like GlusterFS version
    is stored in management address-keyed dicts).

    So far it was available only through ad-hoc mangling of
    various attributes of a GlusterManager instance that
    represents the volume. Now we make it available
    directly from GlusterManager as an attribute.

    Also fix an occurrence of ad-hoc mangling going wrong,
    ie. where a wrongly constructed address is attempted
    to be used as dict key.

    Change-Id: Ic5a96bc99943dda3592372512916257d53f61b80
    Closes-Bug: #1476710