[openstack-dev] [nova] Backwards incompatible API changes

Christopher Yeoh cbkyeoh at gmail.com
Mon Mar 24 05:04:06 UTC 2014


On Fri, 21 Mar 2014 07:57:34 -0600
Chris Friesen <chris.friesen at windriver.com> wrote:
> This is sort of off on a tangent, but one of the things that resulted
> in this being a problem was the fact that if someone creates a
> private flavor and then tries to add access second flavor access call
> will fail because the the tenant already is on the access list.
> 
> Something I was wondering...why do we fail the second call?
> 
> It would make sense to me to just return success without actually
> doing anything since at the end of the operation the tenant has the
> access they requested.

Chances are the current behaviour is a result of what was easiest at the
time, rather than an intentional decision. The low level db call raises
an exception if the tenant is already in the access list, so this was
just mirrored in the API.

Hopefully the increased spec requirements will bring theses sorts of
questions up before new features get merged. Generally speaking though I
think we should try to ensure our API is consistent in how it treats
situations like this. 

> Of course, that'd be another API change...  :)

Yea, its perhaps something we can change for the V3 API, but I think we
should consider it in the context of what the rest of the API does.

Chris



More information about the OpenStack-dev mailing list