clone on busy SolidFire cluster may result in delay for _get_model_info

[John Griffith]

Currently the _get_model_update method in the SolidFire driver tries to gather the volume info immediately after a clone operation is requested. Depending on the size of the volume and how busy the cluster is this operation may actually take a few seconds to complete. As things are currently, we do the check, if nothings returned we just return and empty model_upate.

To address this, add a poll routine for up to a certain number of seconds and raise/report an error if the clone isn't available within the specified period of time.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/22395

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/22395
Committed: http://github.com/openstack/cinder/commit/737468e5c80a35b859dcda29706c88e2976339a3
Submitter: Jenkins
Branch: master

commit 737468e5c80a35b859dcda29706c88e2976339a3
Author: John Griffith <email address hidden>
Date: Tue Feb 19 21:34:01 2013 -0700

    Handle maxclonepervolume/node limits in SF driver.

    The SolidFire cluster has a max simultaneous clones per volume and node
    limt that we weren't accounting for. Add some logic to detect this situation
    and loop until enough simultaneous clones are completed to allow processing
    of the next one in line.

    Fixes bug: 1129810

    Change-Id: Id63c1b142a81df5d9d2c4e39fb821b03ebac63a2

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/folsom
Review: https://review.openstack.org/25964

[Alan Pevec (apevec)]

stable/folsom review was abandoned