On 7/28/2016 11:20 PM, Ghanshyam Mann wrote:
Yes, I also prefer the approach of capping the tests instead of jobs. But along 
with that we might need to make sure the same tests coverage Tempest provides if 
min_microversion is set >2.35 in config.
For example, if we cap the tests (calls nova-network) with max_microversion = 
2.35 then, we might need to implement/modify those tests to start using neutron 
which can be run if config's min_microversion
is set > 2.35.
There are two type of test cases-
    1. Test only tests nova-network APIs - Example: 
https://github.com/openstack/tempest/tree/master/tempest/api/compute/floating_ips
    2. Test testing other scenario using nova-network - Example: 
https://github.com/openstack/tempest/blob/master/tempest/api/compute/servers/test_server_rescue.py

1st case if all ok to cap and leave it skip for >2.35. But for 2nd case, I feel we 
should not leave them skip if config's min_microversion > 2.35 which mean leaving 
those scenario untested(if min_microversion >2.35).
There are 2 options for 2nd case:
  1. Implement  duplicate tests by using the neutron APIs - This will be 
duplicate tests but needed because of testing nova-network till newton EOL.
  2. Or modify those to switch from nova-network to neutron.  - If we do not 
care about nova-network testing even for stable branches where it is not 
deprecated.

I don't like either option, but I'd rather go with #1 for two reasons:

a) Once nova-network is gone these tests wouldn't run ever, so we lose the coverage.

b) nova-network wasn't deprecated in stable branches so I think we need to maintain those tests, so 2.1-2.35 are nova-network, 2.36+ are neutron.

The duplication cost probably wouldn't be all that terrible, we could probably abstract a lot of the common network setup parts away for the compute API tests for things like creating a floating IP.



Thanks
gmann

__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev



--

Thanks,

Matt Riedemann


__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to