HNAS NFS driver allows the creation of cloned volumes using volume
types that are not the same as the ones from the source volume. This
could make cinder information inconsistent with the real volume
location in the backend when we have multiple pools configured.
As the clones are always created in the same pool as the source
volumes, this patch makes the driver accept the creation of cloned
volumes only if they have the same volume type as the source volume.
Closes-Bug: #1660411
Change-Id: I1272a6d27d7f172e77f1260a2acc6fc7c02e2e73
When we create temporary volumes, from another volume or from a
snapshot, we are not storing the cluster_name in the DB, so later on
it's problematic to delete them.
TrivialFix
Closes-Bug: #1660750
Change-Id: I87b73ae709411d21631654ac8d40cd4fe70bccaf
1. While creating a consistency group from a source consistency group
having thinly provisioned volume(s), the target volumes would not
be thinly provisioned because 'tpvv' parameter was not being passed.
Same thing was happening for thinly deduplicated volume(s).
Extracted the optional boolean parameters 'tpvv' and 'tdvv', if present
, from the source volume(s) to be used for corresponding target
volume(s) creation. In case these are not present, default value of
False is used.
Fixed the test case too so that it could accept these addtional two
parameters.
2. Cloning a CG with a bootable volume would create a CG with a
non-bootable volume. This was due to 'bootable' flag of source
volume not being used while creating DB entry for target volume.
Closes-Bug: #1655541
Change-Id: Ia997629a3b2189d1722f87edda5f3989fe79ffdb
They previously didn't have any description in the
driver list due to not having any doctoring for the
classes. This adds a little blurb in...
We also need to bump the driver versions for the O
release.
Change-Id: I237c70e24dcf2513f807e8dcb950d425148c001d
Closes-Bug: #1660479
in the xiv portion of the ibm_storage driver, delete volume
silently failed if the volume was replicated, due to
invalid argument name.
The result was an unmanaged volume left on the storage,
taking up space.
The argument name has been fixed and the volume is cleaned up.
Change-Id: I58bc8f91bdee713dc064547e682f95d9b2cb6cd2
closes-bug: 1660127
When deploying Cinder as an SDS without Glance we have no way to prevent
volume creation from images even when we know they will not succeed.
This patch adds a specific policy so we can prevent this specific
creation action from being accepted. By doing so the user will know
immediately that this is not possible, instead of having to look through
the logs to see that this is not an option.
TrivialFix
Change-Id: Iabc10a1927eea6419dd677a632cfc7d32dc08091
When running Cinder with notifications disabled we are still doing all
the work and calls to Oslo message notifier and it's there, at the very
last moment when it's going to send the message that it checks that
there's no actual extension loaded in the driver manager and skips the
actual send of the data.
This is considerably wasteful considering that for some of the
notifications we are actually querying the DB to get data, for example
volume attachments and glance metadata information when notifying about
volume usage.
This patch proposes short-circuiting notification methods as much as
possible to optimize code execution when Cinder has no notification
transport mechanism configured, as is the case when deployed as a
standalone SDS service.
Closes-Bug: #1660303
Change-Id: I77f655d3ef90088ce71304da5d4ea7b543991e90
When we received feedback from volume replication we removed
six.text_type from a dict instead of the intended
replication_status field. This causes an exception.
Change-Id: Ibb74ec6b5fe52e3ec15515bb180259d64c7a6cf4
Closes-Bug: #1659531
Rebrand and organize the EqualLogic driver under the dell_emc
volume driver folder. The EqualLogic product has gone by PS
Series for some time, so this also changes the name to match.
Config options are kept with the eqlx_ prefix since changing
that would add little value and just create additional work for
operators.
Change-Id: I16bd0ba7755ddfcd5a4edd3e11077725093c4d3d
The previous fix for Live Migration
https://review.openstack.org/#/c/420214/ broke our Live
Migration. When checking for Error message in exception,
it needs to be typecast to six.text_type()
This patch resolves the problem.
Change-Id: I0871fb0f5a63cfe17ebf328e7ce8a2fff34fcdb4
Closes-Bug: #1481953
In the xiv portion of the ibm_storage driver, create_consistencygroup
fails when it receives a Group object instead of a ConsistencyGroup
object. Need to handle both objects, and handle multiple group ids.
Change-Id: I21783db166515f80d83ef0d152d0795f8ca0297d
Closes-bug: 1658782
QoS key value was changed from 'qos_specs' to 'qos_spec' in a
feedback comment in _initial_setup. The side effect of this
was QoS was never enabled. Please refer to
https://review.openstack.org/#/c/307502/
Change-Id: I32a203c1a29e214656f6684268266df589db67bd
Closes-Bug: #1656029
Some test classes mock volume_types.get_volume_type_extra_specs
globally, if other drivers use it and run after it, they will
have problems.
Change-Id: Ie25c35e8c04457a09f95230185586703b1fce83f
Closes-Bug: #1659481
