Error while deleting failed replication backup hides root cause

[Peter Stachowski]

If there is a problem with getting the replication snapshot, the backup is deleted. If the delete fails (for example, the backup is in the wrong state) then the original exception is lost (and more importantly, not logged).

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/131593

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/131593
Committed: https://git.openstack.org/cgit/openstack/trove/commit/?id=b2914b7d3fdfa95385748a70745ac6f4b52e1637
Submitter: Jenkins
Branch: master

commit b2914b7d3fdfa95385748a70745ac6f4b52e1637
Author: Peter Stachowski <email address hidden>
Date: Tue Oct 28 03:11:38 2014 -0400

    Deleting failed replication backup can hide error

    If there is a problem with getting the replication snapshot, the backup
    is deleted. If the delete fails (if, for example, the backup is in the
    wrong state) then the original exception is lost (and more importantly,
    not logged).

    Handling is now in place to trap exceptions while deleting a 'bad'
    replication backup. All exceptions are logged and re-thrown as a
    TroveError.

    The duplicated code around the backup delete has also been collapsed
    into one (nested) exception.

    Change-Id: I858df7fd7eeb79dd1d559c4dd4b2851f7d0e0d98
    Closes-Bug: #1386901