Previously deleting a key in certain qos specs was accomplished via
'update' API. Unfortunately, 'update' isn't able to tell the
difference between setting a key with no value and deleting a key (and
its value).
This change adds an new API 'delete_keys' to qos_specs API extension.
'delete_keys' API allows client to specify a list of to-be-deleted keys
in one single request (batch mode!), which can be handy when removing
multiple keys in a qos specs.
Example URL and request body for 'delete_keys' API:
PUT to http://127.0.0.1:8776/v2/qos-specs/QOS_SPECS_UUID/delete_keys
with body: {'keys': ['foo', 'bar', 'zoo']}
Above example request will result in 'foo', 'bar', 'zoo' key/value
pairs of QOS_SPECS_UUID be marked as deleted in DB. If QOS_SPECS_UUID
doesn't exist, a 404 error will return; if any key in 'foo', 'bar',
'zoo' couldn't be found in QOS_SPECS_UUID, a 400 error will return with
error message telling which key couldn't be found. Note that Cinder
will puke 400 and stop trying the rest once it encounters a
non-existing (or deleted) key amoung the given list of keys.
This change also fixes 'list'/'show' API includes deleted keys in
result.
Jay S. Bryant [Mon, 12 Aug 2013 17:55:48 +0000 (12:55 -0500)]
Add check for qemu-img to image_utils fetch_to_raw
Some platforms, particularly PowerPC systems running RHEL,
do not have qemu-img installed by default and do not
support image formats other than RAW. For these systems,
functions/drivers that use fetch_to_raw() currently fail
because it is assumed that qemu-img will always be installed.
This change updates fetch_to_raw() to function more like the
upload_volume() function which checks the image format and skips
any conversion if the image is already in the RAW format. To
minimize the impact upon platforms that have qemu-img installed
this is being implemented by checking to see if qemu-img is
installed. If it is installed, no changes are made to the flow
through fetch_to_raw. If qemu-img is not available and the image
format is not already RAW, an exception is thrown. Otherwise,
the image is downloaded and we do not progress to the portions of
fetch_to_raw that require qemu-img to be installed.
This commit also updates the test cases that touch this code.
Fixed erroneous force full copy in ceph backup driver
The following steps should not result in a forced full copy:
1. create vol
2. create backup (works fine i.e. uses incremental diff)
3. delete backup
4. create new backup (full copy forced because volume backup-snap exists and is not ignored)
cinder.volume.rpcapi.create_volume does not convert the request_spec to
primitive before casting the request. This makes requests containing non
primitive types to fail. For example:
This will create a new database record and call create_volume on
volume.rpcapi. This will fail because VolumeMetadata won't be
serialized correctly when calling cast. This, however, is not True when
--source-volid is not passed because in such case,
scheduler.rpcpai.create_volume will be called, which converts
request_spec to primitive correctly.
Fix a corner case in associating QoS Specs with Volumt types that was
not handled well: Raise HTTPBadRequest() to client if associate volume
type, which is already associated one qos specs other than specified
qos specs. Previously such operation will proceed without error which
results overwriting original association.
Also refactor DB API volume_type_qos_specs_get() to return qos specs
in consistent dict format as other qos specs related APIs.
Victor Rodionov [Fri, 6 Sep 2013 22:18:48 +0000 (02:18 +0400)]
Fix bug in Nexenta NFS driver _do_create_volume
Fix bug in _do_create_volume method of Nexenta NFS volume driver.
NMS call nms.folder.create doesn't support third argument, that is why
this call changed to nms.folder.create_with_opts.
If cinder-backup service is not enabled, "cinder backup-create"
command fails like the following. As the result, the volume status
has been changed to "backing-up" in spite of not backing-up.
$ cinder backup-create f48aa6ae-4c35-4a6a-a393-5a5cf610945a
ERROR: Service cinder-backup could not be found.
$ cinder list
+--+------------+--------------+------+-------------+----------+-------------+
|ID| Status | Display Name | Size | Volume Type | Bootable | Attached to |
+--+------------+--------------+------+-------------+----------+-------------+
|..| backing-up | vol-test | 1 | None | False | |
+--+------------+--------------+------+-------------+----------+-------------+
$
To avoid this situation, this patch moves the cinder-backup check
before changing a volume status to "backing-up".
John Griffith [Fri, 6 Sep 2013 19:55:39 +0000 (13:55 -0600)]
Do not attempt vg.update on uninitialized vg
It's possible for the scheduler to send the periodic
stats update request *before* the volume service/driver
has finished initializing, resulting in an exception trace
being logged.
This change adds a check to verify that the vg is actually
initialized (completed check_for_setup_error) and it also
removes all of the duplication of each LVM class driver.
To deal with other drivers that may have similar issues,
add initialized member to base driver, that member is
set False on __init__ and is updated by the manager
after running check_for_setup_error. We also skip
update_stats if initialization isn't set True.
Rather than have a separate copy of update_stats in every
driver, we have one in the base LVM class and we set the two
unique variables as member parameters.
Zhongyue Luo [Fri, 6 Sep 2013 04:10:28 +0000 (12:10 +0800)]
Fixes misuse of assertTrue in test scripts
Misuse of assertTrue in places where assertEqual should be used.
If assertTrue is used instead of assertEquals, the test will always pass
if the first argument's boolean cast is True. The second argument
passed to assertTrue() will be regarded as the error message
when the assertion fails. Using assertEqual will actually check
if two arguments are same value.
Zhongyue Luo [Fri, 6 Sep 2013 04:01:26 +0000 (12:01 +0800)]
Utilizes assertIsNotNone
Using assertTrue and the 'is not' operator to test if an element
is not None is too python2.4. Our unit testing framework supports
assertIsNotNone, included from python 2.7, which was created
for these types of tests.
Zhongyue Luo [Fri, 30 Aug 2013 11:50:26 +0000 (20:50 +0900)]
Utilize assertIsInstance
Using assertTrue and the 'isinstance' function to test
if an object is in an instance of some class is too python2.4.
Our unit testing framework supports assertIsInstance which was created
for these types of tests. Let's use assertIsInstance for these tests.
Up till now we only had copy-on-write for cloning from snapshot. This
change optionally allows clone from volume to use copy-on-write
instead of a doing a full copy each time. This should increase speed
and reduce nearterm storage consumtion but could introduce some new
risks e.g. excessively long clone chains and flatten storms. To avoid
this, a new config option has been providedons are provided -
rbd_max_clone_depth - which allows the user to limit the depth of a
chain of clones i.e.
a->b->c->d as opposed to a->b
->c
->d
This will avoid flatten storms by breaking chains as they are formed
and at an early, predefined stage.
A second option - rbd_clone_from_volume_force_copy - allows the user
to use a full copy as before i.e. disable COW for volume clones.
VMDK copy_image_to_volume and copy_volume_to_image
Implemented copy_image_to_volume that creates a new volume backing (vm) and
copies over the vmdk file from the glance image. Only glance images of disk
format 'vmdk' can be used to create a volume using this driver.
Also implemented copy_volume_to_image that creates a new glance image using
the volume's vmdk file. The steps involved are to take a snapshot of volume
vmdk, take a copy of this read only file and upload to glance.
Cleaned up docstrings as pointed out in earlier reviews.
Added timeout to avoid blocking wait on read/write threads and also fixed a
bug to upload image from copy of vmdk instead of from original vmdk during
upload to glance.
Kurt Martin [Wed, 4 Sep 2013 22:08:51 +0000 (15:08 -0700)]
Validate VV Set exists in 3PAR drivers
The 3PAR drivers must first create a volume and then add that
volume to a virtual volume set on the backend for QoS support.
If that predefined virutal volume set does not exists the
volume would not be associated with the correct QoS settings.
This patch will now look to see if the virtaul volume set
exists and if not will delete the volume and notify the user
that the VV Set does not exists.
We need to let folks know where brick is
currently being maintained for the Havana
release and where to file defects.
Once this simple patch lands, we'll pull this into
the Nova patch.
(fix hacking violation in commit message for the 2nd
time on this patch. To prevent it from blocking the
merge queue on feature freeze day.)
PLEASE DON'T +A PATCHES UNTIL THEY PASS JENKINS CHECK
The exc_info is just a tuple (captured in taskflow)
and isn't itself a callable object so we should not
need to use it like one to extract the exception type
and value.
Running pep8 tests in Nova gives a violation
in lvm.py for not importing a module. This
is a simple fix to import the module itself
instead of only importing a function in the
module.
Kun Huang [Wed, 4 Sep 2013 16:36:48 +0000 (00:36 +0800)]
fix inconsistent i18n log message
An inconsistent example:
imessage = _("Invalid metadata") + ": %(reason)s"
This is not consistent i18n usage. In most of cases, we use _() for
whole string instead of leaving something out.
More than inconsistent, some text in leaving strings may be not
translated, for example, ':' in English is different '：' in Chinese.
Eric Harney [Mon, 19 Aug 2013 04:21:54 +0000 (00:21 -0400)]
QEMU-assisted-snapshots for GlusterFS volumes
Coordinate with Nova to create and delete snaphots for
GlusterFS volumes that are attached to VMs.
Cinder is responsible for creating a QCOW2 file which Nova
will activate in the VM's snapshot chain when a snapshot is
created.
When a snapshot is deleted, Cinder will request for Nova to
perform a block commit/rebase operation to logically delete
the snapshot from the QCOW2 chain.
- move to Zadara APIs 13.07
- added support for extend volume
- added support for create/delete snapshot
- added support for create clones from volumes and snaps
- added support for multi-backend
- added volume stats
- added tests
Set vg_thin_pool to pool name instead of pool_path
create_thin_pool is setting vg_thin_pool to the pool path instead of the
pool_name. This makes volumes creation fail when the create_thin_pool
method is called. This happens because create_volume builds the pool
path itself as create_thin_pool does.
Keeping the pool name in vg_thin_pool instead of the path makes more
sense and allows it to be used in other places in the brick. Also, most
commands return both vg_name and pool_name separated.
The Windows service fails due to missing non-blocking IO features
in eventlet. This fix adds a conditional path on Windows to execute
the service accordingly.
Michael Kerrin [Tue, 3 Sep 2013 12:53:34 +0000 (12:53 +0000)]
Fix errors in volume usage audit script
The snapshot_get_active_by_window DB query was returning snapshots
where we couldn't access the related volume object. This was due to
a) the volume not been loaded and the join failing due to detached
session object.
and b) the related volume was deleted so it didn't show return from
the join
Eric Harney [Fri, 19 Jul 2013 14:02:51 +0000 (10:02 -0400)]
New update_snapshot_status API
Adds new snapshot_actions module
Update_snapshot_status: Allows updating of 'state' and
'progress' fields of a snapshot. This is used by Nova
to inform Cinder of its outcome when performing snapshot
operations for attached volumes. Updates are restricted
to a subset of possible start and finish states.
Huawei OceanStor HVS-series enterprise storage system is an optimum
storage platform for next-generation data centers that feature
virtualization, hybrid cloud, simplified IT, and low carbon footprints.
This patch add an iSCSI driver and a FC driver for Huawei HVS storage
system, using REST. We define a common module for both iSCSI driver and
FC driver. The drivers support volume type, QoS.
zhangchao010 [Sat, 31 Aug 2013 02:24:43 +0000 (10:24 +0800)]
Add Fibre Channel drivers for Huawei storage systems
This is the third patch, changes as follows:
1.Add Fibre Channel drivers for huawei OceanStor T series and Dorado
series arrays. Dorado FC driver inherits codes from FC driver of T.
The FC drivers call module ssh_common which has been defined in
the preview patch: https://review.openstack.org/#/c/41721/
2.Add unit test for the changes.
zhangchao010 [Sat, 31 Aug 2013 02:23:35 +0000 (10:23 +0800)]
Refactor huawei Dorado array iSCSI driver
This is the second patch, changes as follows:
1.Add ISCSIDriver for Dorado arrays. The ISCSIDriver inherit from T.
1.Add a common class DoradoCommon for both FC and iSCSI drivers. The common class
inherit from T common for they have many common functions.
2.Add unit test for Dorado drivers.
zhangchao010 [Thu, 29 Aug 2013 02:33:52 +0000 (10:33 +0800)]
Refactor Huawei iSCSI driver
We plan to refactor Huawei iSCSI drivers and add Huawei FC
drivers. For that's a huge change, we break the codes into
three patches:
1.Refactor T iSCSI driver
2.Refactor Dorado iSCSI driver
3.Add FC drivers for both T and Dorado arrays.
This is the first patch, changes as follows:
1.Define a common class for both FC and iSCSI drivers, and
also provide a unified class HuaweiVolumeDriver for users.
The unified driver will call HuaweiTISCSIDriver according to
users' configuration. The HuaweiTISCSIDriver is a subclass of
driver.ISCSIDriver, so it could get good inheritance.
2.Support volume type.
3.Refactor unit test to make it more logic clear and add more
test cases to get higher coverage rate.
Code Review / openstack-build / cinder-build.git / log