Glance v1 APIs were removed in Rocky and tempest master
does not support the Rocky release. If glance v1 APIs
needs to be tested for older release then older Tempest
can be used.
Tempest is removing the Glance v1 APIs tests, config option,
and its service clients and this change is needed for that.
Needed-By: https://review.opendev.org/c/openstack/tempest/+/890592
Change-Id: I0ef67e86730320755e6f642a36f97ab462fe0aad
As tempest.scenario.manager was announced stable interface in Tempest 27.0.0[1]
it can be now reused in plugins.
Barbican tempest plugin still uses nova-network old interface for floating ip
related methods, this patch removes them and use tempests's ones which
already use neutron's interface.[2]
Replaced/Removed methods:
* _default_security_group
* _create_security_group
* _create_loginable_secgroup_rule
* _create_security_group_rule
Etherpad concerning this effort:
https://etherpad.opendev.org/p/tempest-scenario-manager-cleanup
[1] https://docs.openstack.org/releasenotes/tempest/v27.0.0.html#release-notes-27-0-0
[2] bbc9dd34f6
Change-Id: I7605fe11caa8ae5725e0c0583c623ebad73d40da
The openstack can have multiple glance stores deployed/available.
It may be a proper thing to copy newly created signed image into
all the available glance stores so barbican tempest tests can access
image regardless on which compute and storage backend they are ran on
and regardless on which glance store is local.
Additionally there is a nova-compute conf parameter[1] which can even
prevent instances being spawned from image which is not available in local
glance store.
The copy-image would happen only if import_image tempest cong option
is available which indicates glance multistore is available.
[1] https://review.opendev.org/#/c/657078/
Change-Id: I1f0d8be1b237da0c96e820c4b3dca09a83b29752
The repo is Python 3 now, so update hacking to version 3.0 which
supports Python 3.
Fix problems found.
Change-Id: Ia59a23a36ebb8548a20e894000f7342c73012eac
1. Tempest has an option CONF.compute.compute_volume_common_az to make
sure that Volume and Instance are created in common AZ If Availability
zones are used. The paramater is respected when instance is created
because tempest.common.compute.create_test_server() is used but not
respected when volume is created for test purpose. Let's make sure
AZ is set when volume is created if parameter is set.
2. The plain Encryptor provider is not support on ceph backends, the
test test_encrypted_cinder_volumes_cryptsetup fails on:
cinder.exception.VolumeDriverException: Volume driver reported an error:
Provider plain not supported.
This comes from [1] which is called in [2]. The Plain provider will be deprecated
in future completely.
[1] https://opendev.org/openstack/cinder/src/branch/master/cinder/volume/volume_utils.py#L1195
[2] https://opendev.org/openstack/cinder/src/branch/master/cinder/volume/drivers/rbd.py#L877
Change-Id: Ib6290ed1ad5779e4e6fed449f3335dbdada42f0d
* From rocky release projects are consuming the volume featured
enabled v3 API, Since the earlier code has support
for v1, v2 and since api_v3 is used in tempest.conf, it will give
AttributeError: 'Manager' object has no attribute 'volumes_client'.
The above patch fixes the issue.
* Keep cinder v1 api as tempest plugin is branchless, it would avoid
breakage if someone is still consuming it.
Closes-Bug: #1775353
Change-Id: Ia9d070a6033a8192112dbaa1704b1de6838051c5
Cinder volume delete [1] is executed during resource cleanup.
So, if test fails due to some reason then it tries to delete a volume
which is "in-use" state. Detach volume should be executed before
trying to delete the volume. This patch adds nova_detach volume
as a cleanup task.
[1] a6d945d8e6/barbican_tempest_plugin/tests/scenario/manager.py (L229)
Change-Id: I3457bb6fc22a7bb6a32c8d6f4ea5879aeb8d43c5
Related-Bug: #1755485
This should help to isolate failures that are generic from specific test
failures related to barbican. If a test can't pass simple connectivity
check, it's probably broken, and there is no point to attach disks,
because it won't work.
Related-Bug: #1742936
Change-Id: I96cd35392f638c0e5b668f4f3b15cf4cd6114b71
_log_console_output is defined in tempest.scenario.manager.Manager but
the test class inherited from BaseTestCase that doesn't have it. So
whenever a test failure occurs, we don't log console output, which may
be handy when debugging gate issues.
Change-Id: I48701da8714c9e483401fd1ba987e4f2bef5bd23
Related-Bug: #1742936
In tempest, alias 'manager' has been moved to 'os_primary'
in version Pike, and it will be removed in version Queens.
This patch is to replace the usage of 'manager' with 'os_primary'.
For other details, please check [1] and [2]
[1] https://review.openstack.org/#/c/468036/
[2] https://review.openstack.org/#/c/463484/
Change-Id: I522405288452f42824991e9709eec4e57aab2606
This is in response to the mailing list post at:
http://lists.openstack.org/pipermail/openstack-dev/2017-February/112938.html
The tempest.scenario.manager class will be refactored,
so a local copy will be maintained in the
barbican_tempest_plugin repository until that refactor
is complete.
Change-Id: I706e256fe4e224e44532ea0ce2c81495791ec97f