cinder-backup doesn't set a useful fail_reason on backup record when swift is unavailable

[stephen mulcahy]

When a backup is created - if there is a failure, we set the backup to the error state and give a detailed message about the source of the error in the fail_reason field. In the case of swift being unavailable (because it is not in the list of ENABLED_SERVICES in a devstack localrc for example) - a create backup currently logs the message "[Errno 111] ECONNREFUSED" for the fail_reason.

The backup, restore and delete operations in the swift service need to catch socket.error 111's and raise a more explanatory exception for use by the exception handler in the backup manager by raising a new exception such as

classSwiftConnectionFailed(CinderException):
    message = _("Connection to swift failed") + ": %(reason)s"

or similar

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/23166

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/23166
Committed: http://github.com/openstack/cinder/commit/3fc515e77aa8a6bc3218dbd36b8ffbb9e7bf5b97
Submitter: Jenkins
Branch: master

commit 3fc515e77aa8a6bc3218dbd36b8ffbb9e7bf5b97
Author: Stephen Mulcahy <email address hidden>
Date: Thu Feb 28 12:11:39 2013 +0000

    Improved fail_reason for cinder-backup swift connection errors

    Modified swift backup service to catch socket errors when talking to
    swift and raise a specific SwiftConnectionFailed exception in these
    cases. This allows us to provide a more readable error message
    detailing the problem connecting to swift to the end user when they
    view the backup. Also reduced the default number of swift retries
    so devstack environments fail faster - production environments can
    tune these in cinder.conf.

    Fixes bug: 1132791

    Change-Id: Ibca744ea5adcbd31d068ac3d858bde6a4a0c9844