Backup snapshot fail due to not Initializes the driver's client

[huanan]

When backup snapshot, i got the following error:

2016-06-17 22:01:42.163 DEBUG oslo_concurrency.lockutils [^[[01;36mreq-2a4dc4a7-d2ee-40c8-beb5-56951951d970 ^[[00;36m515b0e063767468f89ec2cbe94d3291e c0b0b1dc91814333aa17f2e4e0cc9d34] ^[[01;35mLock "huawei" released by "cinder.volume.drivers.huawei.huawei_driver.initialize_connection" :: held 0.003s^[[00m ^[[00;33mfrom (pid=19372) inner /usr/local/lib/python2.7/dist-packages/oslo_concurrency/lockutils.py:282^[[00m
2016-06-17 22:01:42.283 ERROR oslo_messaging.rpc.server [^[[01;36mreq-2a4dc4a7-d2ee-40c8-beb5-56951951d970 ^[[00;36m515b0e063767468f89ec2cbe94d3291e c0b0b1dc91814333aa17f2e4e0cc9d34] ^[[01;35mException during message handling^[[00m2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00mTraceback (most recent call last):
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/rpc/server.py", line 133, in _process_incoming
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m res = self.dispatcher.dispatch(message)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/rpc/dispatcher.py", line 150, in dispatch
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m return self._do_dispatch(endpoint, method, ctxt, args)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/rpc/dispatcher.py", line 121, in _do_dispatch
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m result = func(ctxt, **new_args)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/backup/manager.py", line 386, in create_backup
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m self._update_backup_error(backup, context, six.text_type(err))
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_utils/excutils.py", line 221, in __exit__
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m self.force_reraise()
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_utils/excutils.py", line 197, in force_reraise
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m six.reraise(self.type_, self.value, self.tb)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/backup/manager.py", line 380, in create_backup
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m self._run_backup(context, backup, volume)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/backup/manager.py", line 416, in _run_backup
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m properties, is_snapshot)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/backup/manager.py", line 893, in _attach_device
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m context, backup_device, properties)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/volume/driver.py", line 984, in _attach_snapshot
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m snapshot, properties)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/volume/drivers/huawei/huawei_driver.py", line 1888, in initialize_connection_snapshot
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m return self.initialize_connection(volume, connector)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/usr/local/lib/python2.7/dist-packages/oslo_concurrency/lockutils.py", line 271, in inner
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m return f(*args, **kwargs)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m File "/opt/stack/cinder/cinder/volume/drivers/huawei/huawei_driver.py", line 1724, in initialize_connection
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m portgroup_id) = self.client.get_iscsi_params(connector)
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00mAttributeError: 'HuaweiISCSIDriver' object has no attribute 'client'
2016-06-17 22:01:42.283 TRACE oslo_messaging.rpc.server ^[[01;35m^[[00m

In Matika backup manage, it use do_setup initail volume driver :
    def _init_volume_driver(self, ctxt, driver):
        LOG.info(_LI("Starting volume driver %(driver_name)s (%(version)s)."),
                 {'driver_name': driver.__class__.__name__,
                  'version': driver.get_version()})
        try:
            driver.do_setup(ctxt)
            driver.check_for_setup_error()
        except Exception:
            LOG.exception(_LE("Error encountered during initialization of "
                              "driver: %(name)s."),
                          {'name': driver.__class__.__name__})
            # we don't want to continue since we failed
            # to initialize the driver correctly.
            return

        driver.set_initialized()

In Newton backup manage, it use import volume manage object initail volume driver by __init__() method:
    def _setup_volume_drivers(self):
        if CONF.enabled_backends:
            for backend in CONF.enabled_backends:
                host = "%s@%s" % (CONF.host, backend)
                mgr = importutils.import_object(CONF.volume_manager,
                                                host=host,
                                                service_name=backend)
                config = mgr.configuration
                backend_name = config.safe_get('volume_backend_name')
                LOG.debug("Registering backend %(backend)s (host=%(host)s "
                          "backend_name=%(backend_name)s).",
                          {'backend': backend, 'host': host,
                           'backend_name': backend_name})
                self.volume_managers[backend] = mgr
        else:
            default = importutils.import_object(CONF.volume_manager)
            LOG.debug("Registering default backend %s.", default)
            self.volume_managers['default'] = default

Most of the vendor driver use do_setup initail volume driver client. So in newton, Those who use do_setup initail volume driver client will not be able to normal use backup services.

[Eric Harney (eharney)]

The backup code is doing the wrong thing here, the contract that people follow when writing volume drivers is that do_setup() will be called before the driver is used.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/331835

[Xing Yang (xing-yang)]

Looks that I missed the "_init_volume_driver" call. I tested with LVM and VNX which do not rely on the do_setup call and didn't catch this problem earlier.

huanan,
Can you please test this patch and see if it works for you now?
https://review.openstack.org/331835

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/331835
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=cbdd0fa1fab24bbb891044b1e4aa56859cab75c4
Submitter: Jenkins
Branch: master

commit cbdd0fa1fab24bbb891044b1e4aa56859cab75c4
Author: xing-yang <email address hidden>
Date: Sat May 21 23:08:25 2016 -0400

    Call _init_volume_driver in backup manager

    When fixing backup using temp snapshot code path, the call to
    _init_volume_driver was missed:
        https://review.openstack.org/#/c/321943/

    For drivers that rely on do_setup to initialize the driver, this
    is a problem because the driver cannot be used without this call.

    This patch fixes the problem by calling _init_volume_driver in
    init_host.

    Change-Id: Ic197046a480dbf8d73a229580e59cd6d489355fa
    Closes-Bug: #1594225

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/mitaka
Review: https://review.openstack.org/334243

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/334243
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=51abc42c6afdfbdc4f706a4f5f657fd86ba3ff2f
Submitter: Jenkins
Branch: stable/mitaka

commit 51abc42c6afdfbdc4f706a4f5f657fd86ba3ff2f
Author: xing-yang <email address hidden>
Date: Sat May 21 23:08:25 2016 -0400

    Call _init_volume_driver in backup manager

    When fixing backup using temp snapshot code path, the call to
    _init_volume_driver was missed:
        https://review.openstack.org/#/c/321943/

    For drivers that rely on do_setup to initialize the driver, this
    is a problem because the driver cannot be used without this call.

    This patch fixes the problem by calling _init_volume_driver in
    init_host.

    Change-Id: Ic197046a480dbf8d73a229580e59cd6d489355fa
    Closes-Bug: #1594225
    (cherry picked from commit cbdd0fa1fab24bbb891044b1e4aa56859cab75c4)

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/cinder 9.0.0.0b2 development milestone.

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/cinder 8.1.0 release.