[
https://issues.apache.org/jira/browse/CLOUDSTACK-9349?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15247705#comment-15247705
]
ASF GitHub Bot commented on CLOUDSTACK-9349:
--------------------------------------------
Github user dmabry commented on a diff in the pull request:
https://github.com/apache/cloudstack/pull/1500#discussion_r60225742
--- Diff: test/integration/component/test_volumes.py ---
@@ -603,7 +603,134 @@ def test_01_volume_attach_detach(self):
"Check the state of VM"
)
except Exception as e:
- self.fail("Exception occuered: %s" % e)
+ self.fail("Exception occurred: %s" % e)
+ return
+
+ @attr(tags=["advanced", "advancedns"])
+ def test_02_root_volume_attach_detach(self):
+ """Test Root Volume attach/detach to VM
+ """
+
+ # Validate the following
+ # 1. Deploy a VM
+ # 2. Check for root volume
+ # 3. Stop VM
+ # 4. Detach root volume
+ # 5. Verify root volume detached
+ # 6. Attach root volume
+ # 7. Start VM
+
+ try:
+ # Check for root volume
+ root_volume_response = Volume.list(
+ self.apiclient,
+ virtualmachineid=self.virtual_machine.id,
+ type='ROOT',
+ listall=True
+ )
+ self.assertNotEqual(
+ root_volume_response,
+ None,
+ "Check if root volume exists in ListVolumes"
+ )
+ self.assertEqual(
+ isinstance(root_volume_response, list),
+ True,
+ "Check list volumes response for valid list"
+ )
+ # Grab the root volume for later use
+ root_volume = root_volume_response[0]
+
+ # Stop VM
+ self.debug("Stopping the VM: %s" % self.virtual_machine.id)
+ self.virtual_machine.stop(self.apiclient)
+
+ # Ensure VM is stopped before detaching the root volume
+ time.sleep(self.services["sleep"])
+
+ vm_response = VirtualMachine.list(
+ self.apiclient,
+ id=self.virtual_machine.id,
+ )
+ vm = vm_response[0]
+ self.assertEqual(
+ vm.state,
+ 'Stopped',
+ "Check the state of VM"
+ )
+
+ # Detach root volume from VM
+ self.virtual_machine.detach_volume(
+ self.apiclient,
+ root_volume
+ )
+
+ # Verify that root disk is gone
+ no_root_volume_response = Volume.list(
+ self.apiclient,
+ virtualmachineid=self.virtual_machine.id,
+ type='ROOT',
+ listall=True
+ )
+ self.assertEqual(
+ no_root_volume_response,
+ None,
+ "Check if root volume exists in ListVolumes"
+ )
+
+ # Attach root volume to VM
+ self.virtual_machine.attach_volume(
+ self.apiclient,
+ root_volume,
+ 0
+ )
+
+ # Check for root volume
+ new_root_volume_response = Volume.list(
+ self.apiclient,
+ virtualmachineid=self.virtual_machine.id,
+ type='ROOT',
+ listall=True
+ )
+ self.assertNotEqual(
+ new_root_volume_response,
+ None,
+ "Check if root volume exists in ListVolumes"
+ )
+ self.assertEqual(
+ isinstance(new_root_volume_response, list),
+ True,
+ "Check list volumes response for valid list"
+ )
+
+ # Start VM
+ self.virtual_machine.start(self.apiclient)
+ # Sleep to ensure that VM is in ready state
+ time.sleep(self.services["sleep"])
+
+ vm_response = VirtualMachine.list(
+ self.apiclient,
+ id=self.virtual_machine.id,
+ )
+ # Verify VM response to check whether VM deployment was
successful
+ self.assertEqual(
+ isinstance(vm_response, list),
+ True,
+ "Check list VM response for valid list"
+ )
+ self.assertNotEqual(
+ len(vm_response),
+ 0,
+ "Check VMs available in List VMs response"
+ )
--- End diff --
The more I look at this, I think I understand why you need 2 different
tests. The first test ensures that returned object is a list, however that
only verifies the type and not the content of the list. The second test
ensures that the returned list isn't empty. I guess it is conceivable that the
VirtualMachine.list could return an empty (0 len) list, which would indeed by a
problem. I would need to look deeper into the cloudstackAPI code to verify
VirtualMachine.list behavior, but that is the only logical reason I can think
of for both tests. I think it is best to leave this second test in here for
now, as I think protects us from an empty list being returned and causing the
assert vm.state= "Running" to fail.
Thoughts?
> Unable to detach root volume when using Hypervisor Type KVM
> -----------------------------------------------------------
>
> Key: CLOUDSTACK-9349
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-9349
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: Volumes
> Affects Versions: 4.5.1, 4.6.2, 4.7.1, 4.8.0, 4.9.0
> Environment: Centos 7
> Reporter: Simon Weller
> Priority: Minor
> Fix For: 4.7.2
>
>
> Back in 4.5, support was added in CLOUDSTACK-6284 for detaching root volumes.
> The original support was meant to work with Xen, VMware and KVM.
> After chatting with fuflo in the Cloudstack irc channel, it was pointed out
> that a constraint was not correctly modified in VolumeApiServiceImpl.java to
> allow the detach to occur when vm.getHypervisorType() == HypervisorType.KVM.
> This is a very useful feature, as it allows us to simulate a snapshot revert
> with Ceph by using createVolume sourced from a snapshot, then detaching and
> reattaching the root volume (new root volume needs to be attached as
> device=0).
> I'm going to propose a PR for this shortly
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
