Open Stack

On 05/02/2014 04:18 PM, Robert Collins wrote:

> I have and remain a huge advocate for trunk being *production ready*,
> because if its not, you're looking at *another* 3-6 months of work
> post release to fix things up. OpenStack is big, with many
> interactions, and if we let things get bad it gets bad very very fast.
> Bad like 'we can't deploy even 30 machines in a Heat cluster' (the
> latest case where something that wasn't tested blew out).

That raises the question of exactly *what* should be production ready. 
Openstack has so many options, so many configuration parameters, so many 
services.

As an example, some people would argue that ceilometer is currently not 
production ready.

I recently stumbled over some stuff because the nova unit tests use 
sqlite and the regex matching is different from the "real" databases. 
Tempest tests would have caught it, but there were no tempest tests for 
that feature.

Has anyone considered having separate validation and release schedules 
for each component?  That way someone making a change in nova wouldn't 
be affected if someone broke cinder, and vice versa.  As it stands, I've 
had trivial patches held up for a week because the automated testing 
system was busted.

Chris