[Libvirt-cim] Re: Test Run Summary (Aug 11 2009): Xen on Red Hat Enterprise Linux Server release 5.3 (Tikanga) with Pegasus

[Kaitlin Rupert]

>>> RASD - 01_verify_rasd_fields.py: FAIL
> The test case do not have any problem.
> I am seeing a peculiar behavior. The EnumerateInstance operation from 
> the provider on the RASD is returning None value in the NetworkName field.
> I commented out the destroy and undefine part of the test case and 
> checked if the wbemcli ein operation on RASD returned proper NetworkName 
> value, which did.
> The rasd_from_vdev() function in Virt_RASD.c is indirectly called via 
> enu_rasds -> _get_rasds().
> The NetworkName value seem to be assigned properly when called from the 
> VSSDC.c vssd_to_rasd() function.
> But when called from EnumInstances() in the Virt_RASD.c the NetworkName 
> is getting reset to null.

Thanks Deepti for providing debug - this helped me track the issue down. 
  The actual error is in the virt_device_dup() function in 
device_parsing.c  I'll follow up with a patch to fix this.


>>> ResourceAllocationFromPool - 01_forward.py: FAIL
> Though the networkpool information is supplied with the Xen domain while 
> creating it.. once created the xml will have the bridge information 
> instead of pool.
> Because of the provider would not have the source networkpool 
> information when queried, this test fails with the following error:
> 
> device_parsing.c(325): No network source defined, leaving blank
> Virt_DevicePool.c(444): Unable to determine pool since no network source 
> defined
> 

This is a consequence of how the libvirt Xen driver works.  When the 
guest is defined, the XML is converted from a "network" type interface 
to a "bridge" type.

-- 
Kaitlin Rupert
IBM Linux Technology Center
kaitlin at linux.vnet.ibm.com