getClusterInfo is very time-consuming on LHN driver

[Jay Lee]

release: master, stable/havana
volume driver: cinder.volume.san.HpSanISCSIDriver

cinder-volume running a _report_driver_status method periodically.
_report_driver_status called get_volume_stats,

self.driver.get_volume_stats(refresh=True)

When using HpSanISCSIDriver,
_get_volume_stats called _update_backend_status.

    def _update_backend_status(self):
        data = {}
        backend_name = self.configuration.safe_get('volume_backend_name')
        data['volume_backend_name'] = backend_name or self.__class__.__name__
        data['driver_version'] = self.VERSION
        data['reserved_percentage'] = 0
        data['storage_protocol'] = 'iSCSI'
        data['vendor_name'] = 'Hewlett-Packard'

        result_xml = self._cliq_run_xml("getClusterInfo", {})

This method call getClusterInfo without searchdepth option.
So It shows every volume, snapshot information, it takes a long long time.
The San storage has more volumes, response time is slower.

In my case, without searchdepth, getClusterInfo output after 546 secs.
with searchdepth=1, output is 5 secs less.

_update_backend_status need "spaceTotal", "unprovisionedSpace" information,
not a volume, snapshot information. so getClusterInfo should be executed with
searchdepth.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/72041

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/72041
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=251705354914d4af761965bf32b0e8fd948951c3
Submitter: Jenkins
Branch: master

commit 251705354914d4af761965bf32b0e8fd948951c3
Author: Jay Lee <email address hidden>
Date: Sat Feb 8 01:43:27 2014 +0000

    Add searchDepth when getClusterInfo called.

    Without searchDepth, getClusterInfo shows every volume and
    snapshot. It's an unused information and takes very long time.

    Change-Id: Id590e3023f4446f01b965baf4e8ac61980101067
    Closes-Bug: 1276388