test_volume_encrypted: Stop creating an image for each instance
An independent image isn't required for these tests and additionally
isn't required when calling create_server as the lack of an image
results in CONF.compute.image_ref being used by
tempest.common.compute.create_test_server [1].
The creation of this image resulted in the nova-ceph-multistore job
breaking as documented in bug #1912607. This broke as we use the job to
provide both our core ceph coverage and slightly more edgey topologies
using Glance multistore and the [workarounds]/ workaround option. This
option blocking the download of rbd images via g-api when n-cpu can't
clone the rbd volumes directly, as caused by the image creation in
test_volume_encrypted.
Iaf6f6e0dbcb25561bf00e969e1964cd30e974e64 is currently skipping any
encryption test in the nova-ceph-multistore job to workaround bug
#1912607 for the time being but we would obviously like to revert that
if possible and expand our coverage.
A follow up change will also propose adding nova-ceph-multistore to the
gate of this plugin to avoid future breakage.
Related-Bug: #1912607
[1] https://opendev.org/openstack/tempest/src/branch/master/tempest/common/compute.py#L108-L109
Change-Id: I50d3b874b7e0676436c434be6c059f221041560f
Lee Yarwood
5 years ago
| 37 | 37 | super(TestEncryptedCinderVolumes, cls).resource_cleanup() |
| 38 | 38 | |
| 39 | 39 | def launch_instance(self): |
| 40 | image = self.image_create() | |
| 41 | 40 | keypair = self.create_keypair() |
| 42 | 41 | |
| 43 | return self.create_server(image_id=image, key_name=keypair['name']) | |
| 42 | return self.create_server(key_name=keypair['name']) | |
| 44 | 43 | |
| 45 | 44 | def attach_detach_volume(self, server, volume): |
| 46 | 45 | attached_volume = self.nova_volume_attach(server, volume) |