Having a Rakefile will allow us to change the actual test commands on
our side rather than relying on changes to the openstack-infra
repository. This should make it a lot faster to change things, but also
easier to test since the jenkins jobs are actually run in this
repository, not the openstack-infra one.
This commit defines the jobs we previously had defined in Jenkins and
uses 'high-level' naming consistently (i.e. lint, style vs. foodcritic,
rubocop).
There is also a :clean task to help with deleting the files generated by
the other jobs.
Also changed foodcritic to run on the source cookbook rather than the
one installed by berks, see
e.g. https://github.com/berkshelf/berkshelf/issues/931#issuecomment-29668369
Change-Id: Iee61f2aecb237102b9caef6e298b0df85c24370b
blueprint: rakefile
There are cases where migrations should not be executed. For example when
upgrading a zone, and the image database is replicated across many zones.
Having the ability to toggle migration execution is useful (and a use-case
we have :))
Change-Id: I0af8f9ce16814b89b72bebfc868f84e48626e101
This is a much clearer way to test LWRPs are executed with
the proper options. The previous approach was hacky, and
relied upon too much stubbing and ruby magik. Also, updated
to berkshelf 2.0 where Berksfile.lock is respected. This
is necessary for berkshelf to lock to the proper cookbook
deps in openstack's CI system.
Change-Id: Ia912efffa137a75352423e7571cb74bca9ab1b5d