[Libvirt-cim] [PATCH] [TEST] Adding verification for DestroySystem() of the domain

[Deepti B Kalakeri]

Kaitlin Rupert wrote:
>>>>
>>>> Testing KVM hypervisor
>>>> --------------------------------------------------------------------
>>>> VirtualSystemManagementService - 02_destroysystem.py: FAIL
>>>> ERROR - CS instance not returned for test_domain.
>>>> ERROR - RequestedState for dom 'test_domain' is not '3'
>>>> Referenced domain `test_domain' does not exist: Domain not found
>
>>>>
>>>> However, the test passes for me if the patch isn't applied.
>>> Yes! This test fails with the changes, this is because the 
>>> DestroySystem() is not just destroying the domain but also 
>>> undefining it.
>>> The VSMS/15*py tc with the new changes also fails for the same 
>>> reason. I am not sure if you got chance to look at the comments to 
>>> "#2 Add try / except to VSMS 15" patch.
>>> Should I make a note of this in the libvirt.org and XFAIL this test ?
>>>
>> Any further comments on this ?
>>
>
> I didn't see your previous email - sorry for the delay here. My 
> original mail was in hopes that you knew why your changes caused the 
> test to fail.
No problem, I was wondering if I had to change the test case.
>
> Last time I tested VSMS 15 with the changes I submitted, I didn't see 
> it fail. I haven't had a chance to revist that patch.
>
> I took a look at this test, and you're right - the reason it's failing 
> is because DestroySystem() is also undefined the guest. So the answer 
> here is to modify the test so that it doesn't call undefine(). Also, 
> make sure the guest isn't in the inactive domain list either.
>
> Not sure why you want to XFAIL the test, as DestroySystem() is doing 
> what is expected.
>
I thought DestroySystem() is equivalent to "virsh destroy" command which 
would just destroy a running domain which was defined and started.


-- 
Thanks and Regards,
Deepti B. Kalakeri
IBM Linux Technology Center
deeptik at linux.vnet.ibm.com