Actually attach a volume to an instance before taking snapshot
In the test test_snapshot_create_with_volume_in_use, the test calls Cinder
"os-attach" for attaching a volume. The "os-attach" to tell Cinder the
volume is attached, but the API doesn't actually attach the volume to an
instance.(Only update volume status in DB)
This is not right test case for taking a snapshot with in-use volume.
In this test, Nova "os-volume_attachment" should be called for volume
attachment.
Also, some Cinder drivers fails assisted snapshot due to this problem.
In order to perform the snapshot properly, this fix is needed.
Closes-Bug #1401110
Change-Id: Ib31e351fe7c3d27824241cf142c213eae287483f
diff --git a/tempest/api/volume/test_volumes_snapshots.py b/tempest/api/volume/test_volumes_snapshots.py
index 21d0a86..b7e9422 100644
--- a/tempest/api/volume/test_volumes_snapshots.py
+++ b/tempest/api/volume/test_volumes_snapshots.py
@@ -70,11 +70,14 @@
self.addCleanup(self.servers_client.delete_server, server['id'])
self.servers_client.wait_for_server_status(server['id'], 'ACTIVE')
mountpoint = '/dev/%s' % CONF.compute.volume_device_name
- _, body = self.volumes_client.attach_volume(
- self.volume_origin['id'], server['id'], mountpoint)
+ _, body = self.servers_client.attach_volume(
+ server['id'], self.volume_origin['id'], mountpoint)
self.volumes_client.wait_for_volume_status(self.volume_origin['id'],
'in-use')
- self.addCleanup(self._detach, self.volume_origin['id'])
+ self.addCleanup(self.volumes_client.wait_for_volume_status,
+ self.volume_origin['id'], 'available')
+ self.addCleanup(self.servers_client.detach_volume, server['id'],
+ self.volume_origin['id'])
# Snapshot a volume even if it's attached to an instance
snapshot = self.create_snapshot(self.volume_origin['id'],
force=True)