Micro release exception for Nova, Glance, Horizon, Keystone

[Martin Pitt]

Chuck Short [2012-06-22 10:48 -0400]:
> We also run a couple of exercises agaisnt the deployment when it
> happens in the lab.

Where "deployment" is "packages from -proposed", or images that were
built using the actual packages from -proposed? We need to ensure that
there have not been any misbuilds or regressions in the packaging
part; i. e. they did not accidentally drop files or put them into a
wrong place, or the postinst fails on upgrade, and similar problems
which upstream tests do not cover.

A deployment test which covers these, together with the continuous
upstream integration testing certainly seems adequate for an MRE to
me. The feedback about that deployment (or other system integration)
test needs to go to the SRU tracking bug (like "update nova to 1.2.3
in precise") to let the SRU team know when an update has been
rubber-stamped as good to go. This did not happen to any of the bugs
in the previous nova SRU and was the main case of why it got stalled.

Thanks,

Martin

-- 
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)