[Libvirt-cim] KVM on Pegasus Test Run Summary for Dec 15 2008

[Deepti B Kalakeri]

Kaitlin Rupert wrote:
>>> FAIL Test Summary:
>>> VirtualSystemManagementService - 14_define_sys_disk.py: FAIL
>> The above tc is failing because of the the missing VirtualDevice 
>> field in the DiskRASD passed to DefineSystem().
>> The net, disk, proc, mem RASD properties passed to the DefineSystem() 
>> in this tc are obtained by the association of AC with 
>> SettingsDefineCapabilities.
>> With the F9 rpm the VirtualDevice field in the DiskRASD is not set 
>> and hence the DefineSystem() throws an error.
>> The tc passes when we include rasd['VirtualDevice'] field to contain 
>> appropriate VirtualDevice value 'hda' in case [KVM] of F9 rpm.
>>
>> I have a doubt here, the changes to throw an error on Missing 
>> VirtualDevice field was included in the modifications made to the 
>> patch with revision 681
>> "Remove the default implied VirtualDevice on DiskRASD", but the 
>> changes included in the F9 rpm are upto 613.
>> Can you let me know upto which changeset does the F9 rpm cover?
>
> This is because the F9 rpm has additional patches applied to it.  The 
> last rpm update we did, we added the DiskRASD patch (along with a few 
> others).
>
> You can tell patches have been applied, because the revision number is 
> 613+ instead of 613.
>
> This makes branching the test difficult because you cannot rely on the 
> revision number alone.
>
> What you can do is check the changeset value.  In const.py, you can 
> designate the change set as the F9 changeset.  This will allow you to 
> branch tests appropriately.
Can you elaborate here more, I did not get your point here.
>
>>> VirtualSystemManagementService - 15_mod_system_settings.py: FAIL
>> The tc fails only with the KVM F9 rpm.
>> The value of AutomaticRecoveryAction is set to 2 only for KVM with F9 
>> rpm, otherwise the value is 3 for Xen and KVM with current sources.
>> I am not able to track if this was expected behavior of setting 
>> AutomaticRecoveryAction to 2 when the F9 rpm was built and was 
>> changed later to be set to 3 for the modifications beyond revision 613.
>> Can you help ?
>
> Looks like this is being caused by a bug in the F9 libvirt rpm.  Do 
> the following to reproduce this issue just using virsh:
>
> 1) Run the test, dump the guest xml.
>
> 2) Modify the guest xml so that <on_crash> is set to restart.
>
> 3) Define the guest using the xml.
>
> 4) Dump the xml of the newly defined guest.  The value for <on_crash> 
> will be destroy, not restart.
>
> I'd add a bug on the wiki page and have this test XFAIL on F9.
>
I have added bug "00010" for the above problem on the wiki, please have 
a look at it.
I have updated the tc and submitted a patch for the same.

Thanks and Regards,
Deepti.