Stop using a forked Rally

[mestery]

Our local Rally changes should be pushed upstream and we should not rely on a fork [1].

[1] http://ovn-scale-test.readthedocs.io/en/latest/install.html

[mestery]

@l8huang @huikang @hzhou8 FYI on this one, I have opened this to track the merge back of our forked rally upstream.

[hzhou8]

@mestery Thanks for tracking this. The ongoing work of unbounding from openstack in Rally seems not yet completed. I saw this spec [1] merged, but some related works has been pending for a long time, e.g. [2], [3].

[1] https://review.openstack.org/#/c/244353/17
[2] https://review.openstack.org/#/c/294521/4
[3] https://review.openstack.org/#/c/294520/6

[huikang]

@hzhou8 I remember months ago you or @l8huang mentioned that rally has been undergoing radical change to support more workload than openstack. This feature accommodates integrating rally-ovs to upstream rally. Is there any update on this? Thanks.

[huikang]

unbound rally from openstack [1]

[1] https://github.com/openstack/rally/blob/master/doc/specs/in-progress/deployment_type.rst

[l8huang]

As I know, the work of unbound rally from openstack is still going on. If we want to stop to use https://github.com/l8huang/rally.git as early as possible, we may make changes on ovn-scale-test to make it depending on upstream rally.

[huikang]

@l8huang Does that mean we need to rebase our change against the latest rally codebase? Thanks.

[l8huang]

We'd better rebase to a release, e.g 5.0, because the master stream may not stable. Any suggests?

[huikang]

@l8huang not yet; it seems the only way to get rid of this issue. Thanks.