[Libvirt-cim] CimTest Report for KVM on F9 13-08-2008

[Deepti B Kalakeri]

Kaitlin Rupert wrote:
>> =======================================================
>> PASS           : 125
>> FAIL           :   1
> > ACUTAL FAILED  :   0
>
> I'd call this an actual fail.  We do see some caching issues with Xen 
> on RHEL 5.2 which causes some tests to fail when they should really 
> pass (will hopefully finish patching those tests soon).
>
> However, we don't have a caching issue with KVM.  So if a test fails 
> but then passes when it's run individually, I'd still consider that a 
> fail.
Yes the caching issue is not seen with KVM. I did try to look at the 
problem. But was not able to find the actual cause for this particular 
problem.
I will try to look into this again.
>
> ---
>
> VirtualSystemManagementService - 06_addresource.py: FAIL
> ERROR   - Got 99:aa:bb:cc:ee:ff, exp 99:aa:bb:cc:ee:ff. Got None, exp 
> default-net51.
>
> It looks like the test tried to add a network device to a guest, and 
> the verification failed.
>
> I'm not sure if this is a test case issue or if there was an actual 
> issue adding the network device.  When I did a bulk run, this test 
> failed due to the following:
>
> ERROR   - Failed to create Virtual Network 'default-net76'
> ERROR   - Unable to create network pool
>
> There's been a number for network pool related issues lately. I'm 
> planning on writing a patch that does the following:
>
> 1) Do some environment setup before any of the tests are called - 
> create a network pool and a disk pool, etc
>
> 2)  Use the network pool and disk pool names in all the test of the 
> tests.
>
> 3) After the tests are finished, destroy the net / disk pools.
>
> The only problem I see with this are tests that need a net / disk pool 
> free environment.  However, this would allow us to create a guest 
> without having to worry about also creating a network pool (and then 
> subsequently cleaning up that pool properly).
Yes doing this will be good.
As of now all of the tc which uses the vxml define, destroy etc methods 
take care of the network pool, the only need is to take care of the 
diskpool.
We need to look into the test case which are calling the 
create_netpool_conf()and also using the vxml define etc methods since 
these might be the tc that might fail intermittently.

Thanks and Regards,
Deepti.